Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

Dr Anne Marie Thow Prof Stephen Jan Ms Erica Reeve Ms Sarah Mounsey

FAO Consultants

FOOD AND AGRICULTURE ORGANIZATION OF THE UNITED NATIONS Apia, 2019

Required citation: Thow, A.M., Jan, S., Reeve, E. and Mounsey, S. 2019. Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji. Rome, FAO. 80 pp. Licence: CC BY-NC-SA 3.0 IGO.

The designations employed and the presentation of material in this information product do not imply the expression of any opinion whatsoever on the part of the Food and Agriculture Organization of the United Nations (FAO) concerning the legal or development status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. The mention of specific companies or products of manufacturers, whether or not these have been patented, does not imply that these have been endorsed or recommended by FAO in preference to others of a similar nature that are not mentioned.

The views expressed in this information product are those of the author(s) and do not necessarily reflect the views or policies of FAO.

ISBN 978-92-5-131254-4 © FAO, 2019

Some rights reserved. This work is made available under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 IGO licence (CC BY-NC-SA 3.0 IGO; https://creativecommons.org/licenses/by-nc-sa/3.0/igo/legalcode/legalcode).

Under the terms of this licence, this work may be copied, redistributed and adapted for non-commercial purposes, provided that the work is appropriately cited. In any use of this work, there should be no suggestion that FAO endorses any specific organization, products or services. The use of the FAO logo is not permitted. If the work is adapted, then it must be licensed under the same or equivalent Creative Commons licence. If a translation of this work is created, it must include the following disclaimer along with the required citation: “This translation was not created by the Food and Agriculture Organization of the United Nations (FAO). FAO is not responsible for the content or accuracy of this translation. The original [Language] edition shall be the authoritative edition.

Disputes arising under the licence that cannot be settled amicably will be resolved by mediation and arbitration as described in Article 8 of the licence except as otherwise provided herein. The applicable mediation rules will be the mediation rules of the World Intellectual Property Organization http://www.wipo.int/amc/en/mediation/rules and any arbitration will be conducted in accordance with the Arbitration Rules of the United Nations Commission on International Trade Law (UNCITRAL).

Third-party materials. Users wishing to reuse material from this work that is attributed to a third party, such as tables, figures or images, are responsible for determining whether permission is needed for that reuse and for obtaining permission from the copyright holder. The risk of claims resulting from infringement of any third-party-owned component in the work rests solely with the user.

Sales, rights and licensing. FAO information products are available on the FAO website (www.fao.org/publications) and can be purchased through [email protected]. Requests for commercial use should be submitted via: www.fao.org/contact-us/licence-request. Queries regarding rights and licensing should be submitted to: [email protected].

Contents

Executive summary ...... v Structure of the Report ...... vii Background and context ...... ix Part 1: Overview of the literature on pricing policies and their use by developed and developing countries ...... 1 Effect on consumption and health ...... 1 Impact on, and use of, revenue ...... 3 Part 2: Establish evidence-based non-discriminatory criteria for nutrition policy in Fiji: A nutrient profiling model ...... 5 A: Selection of appropriate index nutrients ...... 6 Identifying foods and nutrients associated with NCD risk...... 6 Consumption of foods and nutrients associated with NCD risk in Fiji ...... 8 In Fiji, which foods contain nutrients linked to diet-related NCDs in significant amounts? ...... 14 Summary: Foods and Nutrients to target in Fiji...... 18 Step B: Consideration of the most appropriate choice of base (e.g. per 100g, per 100kJ) . 18 Step C: Identification of most appropriate type of model, in relation to intended use of model ...... 19 Step D: Identification of the most appropriate and relevant nutrient thresholds ...... 19 Groups to be exempted from thresholds in the model ...... 19 Step E: Validation and final nutrient thresholds for Fiji ...... 24 Final proposed Nutrient Profiling Model for Fiji ...... 25 Part 3: Identify fiscal policy and priority complimentary measures for incentivizing access to affordable healthy foods in Fiji ...... 26 Review of policy context ...... 26 Policy implications of the model ...... 27 Recommendations for fiscal policy and complementary measures ...... 29 Fiscal policy measures ...... 29 Complementary policy measures targeting discretionary food consumption ...... 29 Complementary policy measures to promote healthy foods ...... 30 Part 4: Model the impact of a science-based health excise on consumption, nutrition and revenue ...... 31 Price elasticity estimates, based on international evidence ...... 31 Price elasticity values for discretionary foods in Fiji ...... 31 Logic models and assumptions ...... 32 Logic models for the impact of the tax ...... 32 Assumptions for modelling ...... 33 Ranges for estimates of impact ...... 33 Estimated impact on consumption ...... 34 Confectionary ...... 35 Cakes, sweet biscuits and baked goods ...... 36 Savoury snacks ...... 36 Beverages ...... 36 Edible ices ...... 36 Instant noodles ...... 39

iii

Estimated impact on health ...... 39 Estimated impact on revenue ...... 40 Estimated impact on household expenditure ...... 41 Conclusion and recommendations ...... 44 Appendix 1: Detailed method for developing nutrient profiling criteria ...... 46 Method used for developing a food classification system for Fiji ...... 52 Key data sources for analysis of nutrient consumption, availability and price ...... 52 Appendix 2: Detailed tables for discretionary foods not meeting thresholds ...... 60 Appendix 3: Summary of price data used in modeling ...... 67 ...... 70

Tables Table 1: Detailed food categories ...... 5 Table 2: Average daily nutrient intake for Fijians over 12 years of age, compared to recommendations ...... 8 Table 3: Top contributors to intakes of calories, fat and sodium, by detailed food group (percentage contribution to total intake) ...... 11 Table 4: Top contributors to intakes of Vitamin A, Vitamin C and Fibre, by detailed food group (percentage contribution percent) ...... 13 Table 5: Energy, fat, sugar and sodium density of packaged and fresh foods in Fiji...... 17 Table 6: Mean nutrient composition of foods and categories being targeted by Fiji’s nutrient profiling model ...... 20 Table 7: Summary of considerations in developing nutrient profiling model for Fiji ...... 21 Table 8: Nutrient profiling model with thresholds proposed for Fiji...... 25 Table 9: Estimated annual revenue generated (FJD) ...... 41 Table 10: Estimated impact on annual individual and total household expenditure (FJD) .... 42 Table 11 Estimated change in weekly expenditure as a percentage of average income across taxes and substitutions ...... 43 Table 12: Summary of types of models for nutrient profiling ...... 56

Figures Figure 1: Percent (%) change in total dietary consumption of nutrients of concern, across tax scenarios (20%, 50% and 100% taxes) and substitutions ...... 34

Executive summary

The aim of this report is to establish an evidence base for the application of food and beverage taxes and complementary measures to encourage dietary substitution towards healthier, local food products in Fiji.

The Government of Fiji has identified action on nutrition and on diet-related Non- Communicable Diseases (NCDs) as a policy priority. The population of Fiji is experiencing a triple burden of malnutrition, due to the rapid dietary transition of the past 5 decades. Consumption of healthy traditional foods such as fish and seafood, staple root crops, coconuts and wild plants has decreased, while consumption of cereals and refined sugar has increased dramatically. In 2011, 85 percent of people reported insufficient consumption of fruit and vegetables, and in 2015 around 60 percent of school age children reported daily consumption of carbonated beverages. The health implications of this transition have been significant. Around half of children under 5 years, and one third of women of child-bearing age suffer from iron deficiency anaemia. The prevalence of hypertension and obesity among adults is more than 30 percent, and one third of teenagers are overweight. In 2011, around 16 percent of adults were found to have diabetes. These high rates of NCDs are associated with significant social and economic costs, including lost productivity and major demands on health care services.

A significant contributor to high salt, sugar and fat intakes are discretionary foods, such as confectionary, snacks and sweet beverages. These foods contribute an average of only 9 percent of daily food intake, but a much higher proportion of nutrients of concern: 16 percent of calories, 25 percent of fat and 20 percent of salt. As a result, reductions in consumption of these products can have a substantial effect on consumptions of nutrients associated with NCD risk. Many of these foods are relatively inexpensive for the significant contribution they make to intakes of energy and nutrients of concern, and thus appropriate targets of taxation. The analysis presented here indicates the application of significant excise taxes (20 — 50 percent) on discretionary foods not meeting nutrient profiling criteria.

We base our recommendations for taxation on a Fiji-specific nutrient profiling model, which sets thresholds for energy, sugar, sodium, fat, and saturated fat across food categories, in line with the Fiji Dietary Guidelines (see p.35 of this report).

Globally, there is growing interest in the application of fiscal policies to create incentives for the consumption of healthier (rather than unhealthy) foods, in order to accelerate action on malnutrition in all its forms, and NCDs This is reflected in the Global Action Plan for Small Island Developing States, and the Fiji National Food Security and Nutrition Policy and Plan of Action. Internationally, the strongest evidence is for the effectiveness of targeted excise taxes on sugar sweetened beverages. There is also opportunity to increase the impact of taxation through investing revenue strategically (e.g. in Berkeley, health promotion programs designed to improve nutrition and decrease consumption of sugary drinks). The literature indicates that significant taxes, for example, 50 percent, would have a meaningful impact on diets and health.

v

Based on the analysis in this report, the specific discretionary food categories recommended to be taxed are:  Confectionary and sugar  Beverages (sweetened drinks, juice, milks)  Edible ices  Cakes, sweet bakery and biscuits  Savoury snacks, including instant noodles

There is opportunity for the revenue generated by taxation to support implementation of complementary measures recommended in the Food Security and Nutrition Policy and Action Plan:  Additional incentives for reformulation, such as nutrient targets for salt, fat and sugar in processed foods, based on the nutrient profiling.  Complementary social marketing, such as a social marketing and school-based campaign that de-normalises consumption of discretionary foods would enhance the effect of the tax, promote public awareness and support for the tax, and further promote reductions in discretionary salt and sugar consumption.  Restrictions on marketing of discretionary foods, to enhance the existing the draft regulation on marketing of foods and beverages to children  Financial support for measures that target healthy food affordability and availability, such as healthy food subsidies.

The estimates for decreased consumption due to taxation of discretionary foods (with ranges due to different assumptions of substitution) were:  20 percent tax: reduction of 2 — 4 percent for calories, 5 — 7 percent for fat and 4— 6 percent for salt.  50 percent tax: reduction of 5 — 10 percent for calories, 7 — 11 percent for fat and 9 — 14 percent for salt

The decreases in intakes of nutrients of concern that are estimated here are associated with reduced risk of diabetes, cardiovascular disease and overweight. In particular, for those aged 60—69 years a SBP reduction of 1 mm Hg has been estimated to reduce risk of stroke mortality by 4 percent and 3 percent mortality from ischaemic heart disease or other vascular causes (Lewington S, 2002).

The total annual revenue generated from the proposed 20 percent or 50 percent tax would equal $FJD51.4 and $FJD74.3 million, respectively. However, this analysis suggests that fiscal policy interventions on discretionary foods are unlikely to represent a significant financial burden to households, due to decreases in consumption of discretionary foods. Our modelling for all scenarios show either neutral effects or reductions in average household spending.

Structure of the Report

In Part 1, we review the global evidence for the use of food and beverage taxes to encourage dietary substitution towards healthier, local food products.

In Part 2, we define an evidence-based criteria (“Nutrient profiling model”) to identify target food and beverage products which contribute to NCD risk in Fiji, which can be used as a basis to develop more targeted policy measures to reduce unhealthy consumption.

In Part 3, we analyse the existing policy context and the policy implications of the nutrient profiling model.

In Part 4, we model the estimated impact of a tax on discretionary foods at 20 percent, 50 percent and 100 percent on consumption, health, revenue and household expenditure

vii

viii

Background and context The Republic of Fiji (Fiji) is a small island developing state in Melanesia, in the South Pacific. The country is comprised of over 330 islands, and the population is approximately 900 000. One of the most significant health issues for Fiji is the extremely high rates of non- communicable diseases (NCDs). Non-Communicable Diseases refer to a group of non- infectious chronic diseases including diabetes, cancer, cardiovascular disease and respiratory disease. There is convincing evidence linking dietary factors to diabetes, cardiovascular disease and some cancers, and the effects of these dietary patterns on NCD incidence are mediated by four key metabolic changes: overweight and obesity, elevated blood pressure, hyperlipidemia and hyperglycemia (Alberti KGM et al, 2005).

NCD risk factor surveys undertaken in Fiji in 2002 and 2011 demonstrated worrying trends in relation to the prevalence of diet-related NCD risk factors. Prevalence of hypertension increased by 50 percent over the 9 years, increasing from 19 percent (WHO. Fiji STEPS Survey) to 31 percent (WHO, Fiji NCD Risk Factors STEPS Report), while rates of obesity have trebled to affect 35 percent of adults. In 2011, around 16 percent of adults were found to have diabetes, with around one-third of Fijians at a high-risk of developing an NCD. Young people in Fiji are similarly demonstrating diet-related risk factors, a 2015 school health survey showed that between 22 percent and 37 percent of students aged between 13—17 years were overweight (WHO, Fiji Global School-based Student Health Survey)

NCDs are the leading cause of death worldwide, responsible for over 38 million deaths worldwide every year (WHO, Global status report on NCD). NCDs are thought to cause a staggering 70 percent of all deaths in Pacific Island Countries (WHO, NCD: Country Profiles). Globally, NCDs represent the most serious threat to social, economic and environmental prosperity. Pacific Island countries, many of which are coping with multiple burdens of NCDS, communicable disease and environmental susceptibilities, are particularly vulnerable to the costs of concomitant lost productivity and demands on health care services. This vulnerability is in part due to their fragile economies, which are dependent on aid and remittances and hampered by recurrent natural disasters (Anderson, I, 2013) (Mavoa HM and McCabe M, 2008).

Like most other Pacific Island Countries, Fiji has undergone a rapid dietary transition over the past 5 decades (Hughes RG, 2003). A traditional pattern of eating reportedly consisted of fish and seafood, a range of root crops, coconuts and plantains, a hybrid mix of wild plants and animals and marine greens, with legumes introduced by Indians in the late 1800’s (Jansen A et al, 1990). Fiji’s National Nutrition Survey (NNS) showed that by 2004, major sources of energy were cereal (bread and flour products, rice and roti, 34 percent), and root crops (20 percent). Sugar was being consumed daily by over 95 percent of households. A number of nutrients were being consumed at levels below (Fibre, Iron, Zinc) or significantly below (Vitamin A and Calcium) recommended levels (National Nutrition Centre. Fiji National Nutrition Survey 2004), and rates of anaemia were particularly concerning, with anaemia prevalent in up to 50 percent of children under 5 years, and between 33 percent and 42 percent in females between 12 and 44 years. Data regarding healthy food consumption from the 2014 National Nutrition Survey (NNS) (National Food and Nutrition Centre, NNS Data) suggest that healthy food consumption may have continued to decline. Only 10 percent of respondents consumed breadfruit daily, and 24 percent cassava. In contrast, 50 percent of

ix

respondents consumed rice daily and 43 percent roti, and daily consumption of instant noodles increased from around 10 percent in 2004 to around 15 percent in 2014. Consumption of vegetables has remained low, and for some vegetables has declined. A review of import data showed that imports of ‘less healthy foods’ increased significantly between 1980 and 2010 (Ravuvu A et al, 2017), in particular white rice, edible oils and sugar. The 2011 STEPs survey showed that 85 percent of people were not consuming enough fruit and vegetables (WHO, Fiji NCD Risk Factors STEPS Report). Among adults, underweight declined from 6 percent to 4 percent between 2004 and 2014, according to the NNS. The prevalence of overweight remained constant, at around 32 percent. However, the prevalence of obesity increased markedly, from 24 percent in 2004 to 32 percent in 2014.

Among children, indicators of undernutrition did not decline between 2004 and 2014. The prevalence of wasting remained at 8 percent and underweight at 7 percent in children under 5 years of age, and stunting at around 10 percent. Stunting among children aged 5 — 17 years declined slightly to 4.5 percent in 2014, while underweight prevalence increased from 7 — 10 percent among children aged 5—14 (Using BMI for age), However, the prevalence of overweight among children and adolescents increased from 6 percent in 2004, to 8 percent (5-14 years) and 12 percent (15—18 years) in 2014. The 2015 Fiji School Health survey found that between 59 percent and 63 percent of youths consumed carbonated beverages almost every day.

Pacific Island governments have committed to taking policy action to prevent and control the escalating burden of NCDs. Most recently, the NCD Roadmap report identified the need for policies to reduce consumption of food and drink products directly linked to obesity, heart disease and diabetes in the Pacific, especially salt and sugary drinks (NCD Roadmap Report, 2014). The policy recommendations in the Roadmap include settings-based promotion of healthy diets, including in schools and workplaces; social marketing of healthy local foods; agricultural and related interventions to increase production and accessibility of healthy, local staples, fruits, vegetables and fish; restricting advertising of unhealthy foods; taxation of unhealthy foods; restricting retail of unhealthy foods; and interventions to reduce salt, fat and sugar in processed foods.

This reflects wider commitment by Pacific Island Country Health Ministers to the Healthy Islands concept, and the “Pacific Framework for the Prevention and Control of Non- communicable Diseases” (Honiara outcome, 2011). This framework includes addressing NCDs through macroeconomic decisions and policy interventions, as well as lifestyle and clinical interventions, and encourages countries to implement cost-effective interventions for prevention and management of NCD priority interventions.

Taking action on nutrition and NCDs is a priority for the Government of Fiji, as articulated in its 2017 National Development Plan (National Development Plan: Transforming Fiji, 2017). The 2017 Fiji Food Security and Nutrition Policy and Plan of Action identify the need to establish the nutrition impact, and impact on revenue, of food and beverage taxes.

The aim of this report is to establish an evidence base for the application of food and beverage taxes and complementary measures to encourage dietary substitution towards healthier, local food products in Fiji. This report reviews the global evidence for the use of

x

food and beverage taxes to encourage dietary substitution towards healthier, local food products; defines evidence-based criteria to identify target food and beverage products; assesses the potential impact of the proposed health on consumption, nutrition and government revenue; and identifies complimentary measure most effective at improving access to an affordable supply of nutritious, local foods to encourage dietary substitution.

xi

Part 1: Overview of the literature on pricing policies and their use by developed and developing countries

Part 1: Overview of the literature on pricing policies and their use by developed and developing countries

Consumption of products high in energy, saturated fat, trans-fat, sugar and salt is a risk factor for overweight and obesity and some NCDs (Ezzati M and Riboli E., 2017). A key component of policy interventions to prevent diet-related NCDs is thus reductions in consumption of such products, and concomitant increases in consumption of healthy foods, such as fruit, vegetables, legumes, nuts and fish (Mozaffarian D, 2016). The rationale for the use of fiscal policy tools is primarily that they target consumer decision making at the point of purchase, through price incentives, to decrease consumption of such products (and/or increase consumption of healthier products) (Hawkes C et al, 2015 ).

The evidence for the effectiveness of fiscal policy measures as part of a comprehensive policy approach is NCD prevention is growing (WHO, Using price polices to promote healthier diets). Information on the impact of fiscal policy intervention comes from a range of sources. As countries have begun to implement such interventions, there have been a growing number of evaluations of the impact of actual interventions. Modelling and simulation studies are particularly helpful for forecasting potential outcomes in real-world settings for use in policy design. Randomised controlled trials provide detailed insights into consumer behaviour in various settings. (Thow et al, Nutrition Reviews) for a detailed description of the strengths and limitations of different study designs for assessing impact of fiscal policy intervention).

Effect on consumption and health The strongest and most consistent effect of taxation in reducing unhealthy consumption is seen for sugar-sweetened beverage (SSB) taxation. Sugar sweetened beverage consumption is associated with obesity and diabetes, as well as being a contributor to the high global prevalence of dental caries (Schwendicke et al, 2016). Reviews of the impact of SSB taxes (mainly studies from high income countries) have consistently found an approximately proportional impact of taxation on consumption, for taxes over 10 percent, and the greatest impact from taxes of 20—50 percent (Thow et al, 2008) (Backholer et al, 2016). They have also indicated health benefits of small but meaningful reductions in BMI, diabetes and dental caries. However, recent studies from low and middle income countries have indicated consistent effects.

Mexico and Chile have both recently increased taxes on SSBs – Mexico through a 1 peso per litre excise tax (approximately 10 percent), and Chile through increasing sales and import taxes from 13 percent to 18 percent. Analyses of the impact of these taxes have found a decrease in consumption of SSBs proportional to the size of the tax (around 7 percent in both countries), and an increase in consumption of drinking water (Colchero etal, 2016) (Guerrero-Lopez et al, 2017) Modelled studies from other low and middle income countries have supported these findings, and also indicated meaningful health benefits associated with taxes on soft drinks. A 20 percent soft drink tax in India was estimated to potentially

1 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji reduce overweight and obesity prevalence by 3.0 percent and type 2 diabetes incidence by 1.6 percent over the period 2014–2023, taking into account current trends towards increasing soft drink consumption in India (Basu et al, 2014). Similarly, modelling the impact of a 20 percent soft drink tax in South Africa indicated reduced energy intake by about 36kJ per person per day (95 percent CI: 9 to 68kJ) (Manyema et al, 2014). This was estimated to reduce obesity prevalence by 2-3 percent.

More detailed data from high income countries have also highlighted a number of potential differential impacts on consumption, including greater effects among sub-populations consuming high levels of soft drink. A tax implemented in the US city of Berkley ($0.01/oz) resulted in a 21 percent reduction in soft drink consumption in low-income neighbourhoods, and a substantial increase in consumption of plain water (Fable, 2016). Another US modelling study predicted that between 2015-2025, a nationally applied excise tax of $0.01 per ounce of SSB could reduce obesity prevalence by 1 percent among adults and 1.4 percent among youth, saving over 100 000 disability-adjusted life years from obesity-related morbidity (Gortmaker et al, 2015). A recent study based on UK data (Briggs et al, 2017) found that reductions in response to an SSB tax would be greatest amongst 11—18 year- olds who consume the largest volumes of SSBs. It also concluded that the tax could reduce the prevalence of obesity by up to 0.5 percent, and diabetes incidence by up to 18 percent. Studies from the UK and Germany have also indicated that SSB taxation can reduce caries development, especially in younger men and those on low income (Schwendicke et al, 2017).

Impact of nutrient-based taxes on consumption Nutrient-based taxes include those imposed on specific nutrients (e.g. a ‘fat tax’) and those imposed on the basis of nutrient profiling (for example, a tax on energy-dense, nutrient- poor food, or a subsidy for healthy foods).

A review of modelling studies with various nutrient-based scenarios by Thow et al (Thow et al, 2014) indicated a small but relatively positive effect on consumption of target nutrients. Modelled taxes ranging between 5 percent to 40 percent on individual nutrients (ie. fat, sugar, salt) could reduce consumption of saturated fat by 0—3 percent, sugar/sweet foods by up to 23 percent and sodium by 6 percent however the review also highlights the unintended substitution effect, particularly of the fat tax with increased salt consumption and decreased fruit and vegetable consumption. In contrast, nutrient-profiling taxes are less likely to have this unintended consequence as the target foods are based on the complete nutrient composition of the food, not just a single nutrient. The review found that taxes on ‘unhealthy’ food profiles ranging from 10 percent to 50 percent had, overall, a decrease in purchase and calories consumed but in one study a decrease in healthy fruit and vegetable consumption was observed (reversed when combined with a subsidy on fruit and vegetables).

Targeted nutrient taxes implemented in Europe have been found to have positive impacts on diets. The first ‘fat-tax’ was implemented in Denmark and remained effective between October 2011 and January 2013. The tax targeted meat, dairy, animal fat, oils, margarine and butter blends as well as composed foods containing these products at approximately USD2.9 per kilogram of saturated fat (excluding VAT) (Bodker et al 2015). This led, amongst other price rises, to a 13—16 percent increase for high fat minced beef and cream products

2 Part 1: Overview of the literature on pricing policies and their use by developed and developing countries

(Jensen et al, 2015). The tax resulted in a 4.0 percent reduction in saturated fat intake and increased vegetable consumption however salt consumption increased for most individuals (Smed et al, 2016). Finland has been active in applying various excise taxes on sweets and ice cream since 1926 and currently has a 0.95 per kg excise tax on sweets and ice-cream. Two modelling studies found this tax has reduced consumption of these foods by 23 percent (Thow et al, 2014). In a bid to promote healthier consumption in the Hungarian diet, taxes were applied in 2011 to foods above particular thresholds in sugar and salt (eg. chocolate, salted snacks, sweets, as well as SSBs) (WHO, Using price polices to promote healthier diets). Sales of products subject to the taxes have fallen by 27 percent with a 20—35 percent decrease in consumption observed. This has also resulted in considerable product reformulation by manufacturers.

A recent study based on data from New Zealand indicated that a 20 percent tax on major food contributors to salt and saturated fat intakes, combined with a subsidy of 20 percent on fruit and vegetables, could have overall benefits for diets and health. In particular, through reducing daily purchases of fat and sodium by 5 percent and 7 percent, respectively, increasing fruit and vegetable purchases by 12 percent and 18 percent and reducing population energy purchases by around 5 percent (Ni Mhurchu et al, 2015).

Impact of healthy food subsidies on consumption Reviews indicate that healthy food subsidies can be effective in increasing consumption of the target food however, they may also increase overall caloric intake (Thow et al, 2014). As noted above, subsidies can also be used in combination with taxes to help compensate for the potential impact of taxes on household budgets, and provide an additional incentive for changes to consumption.

For example, a study of the impact of the ongoing ‘Healthy Food benefit’ (Private Health Insurance program) in South Africa found a significant increase in the consumption of vegetables, fruit, starchy, fibre-rich foods, fat-free dairy products, lean protein, legumes and healthy fats and oils. It also reported a statistically significant (p=<0.001) relationship between a 25 percent discount and the consumption of the healthy food choices and a lower BMI (An et al, 2013).

Research on subsidies indicates that they are effective in increasing consumption, and likely to be more cost-effective than education or social marketing interventions despite their additional cost. Recent studies from Australia and New Zealand have shown that price reductions of around 20 percent can have meaningful effects on consumption of fruit and vegetables, and resultant measures of population health (Olstad et al, 2010).

Impact on, and use of, revenue The direct impact of a health-related (food and/or beverage) excise on government revenues is determined by both the size of the change to the tax rate, and the impact on production and consumption of the taxed good (elasticity of demand). To maximise the effect on consumption, a relatively large tax on a relatively elastic food or beverage is required; conversely, to maximise revenue, a smaller tax on a relatively inelastic food or beverage is required. Andreyeva and colleagues estimated that a ‘penny per ounce’

3 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

(approximately 20 percent) tax on soft drinks in the United States of America could generate new tax revenue of $79 billion over 2010—2015 (Andreyeva et al, 2011).

Since most food and beverage products are relatively inelastic (PEE between 0 and 1), applying taxes to specific foods is likely to result in both changes in consumption and the generation of revenue. For example, despite the reductions in consumption reported above, the soft drink tax in Mexico generated $1.3 billion in revenue in 2014 (Kilpatrick, 2015). (Although, it should be noted that Mexicans consume the highest volume of soft drinks, per capita, in the world).

Similarly, the sugar taxes implemented by Nauru and French Polynesia both also exceeded revenue expectations: the anticipated revenue from the tax in Nauru was AUD80 000 (approximately USD67 400), but after the first year of implementation this budget estimate was increased to AUD240 000 (approximately USD200 000); in French Polynesia, the revenue raised from the production tax was around 1 billion CFP/year (USD10 million), and from the import tax was around 350 million CFP/year in 2008 ((USD4.2 million).. The excise tax on soft drinks implemented by the Government in Samoa (at the time, 0.30 tala/litre) generated 9,392,787 tala (approximately USD3.5 million) between 2003 – 2007 from the domestic excise, while the import excise generated 196 238 T in 2005, 237 167 T in 2006 and 453 542 T (approximately USD170 000) in 2007 (Thow et al, 2011).

A common concern raised by industry is the effect of taxes on employment, due to job losses as the result of decreased profitability of the affected sectors of the food industry. A recent analysis by Powell and colleagues for two states in the United States of America indicates that a sugar-sweetened beverage tax could actually increase employment overall, if economic modeling takes into account: 1) increased consumption of other (non- sweetened) beverages (often produced by the same companies), 2) the increases in jobs created elsewhere in the economy as consumers reallocate their spending to non-beverage goods and services, and 3) increased government-related employment resulting from increased revenue (Powell et al, 2014).

There is also opportunity to increase the impact of taxation through investing a component of the revenue strategically. In Berkeley, revenue has supported health promotion programs designed to improve nutrition and decrease consumption of sugary drinks, Philadelphia’s tax on soft drinks was linked to funding for community services such as preschools and parks, and in Mexico, it is planned to use a portion of tax revenue to support the provision of safe drinking water (Roache et al, 2017). French Polynesia established a Health Promotion foundation with the revenue from their tax on products containing high levels of sugar (Thow et al, 2011).

4 Part 2: Nutrient profiling model

Part 2: Establish evidence-based non-discriminatory criteria for nutrition policy in Fiji: A nutrient profiling model

Nutrient profiling of foods is ‘the science of categorising foods based on their nutritional composition, for reasons related to preventing disease and promoting health’ (WHO, Nutrient profiling, 2010). Nutrient profiling enables evaluation of the nutritional quality of a food. It can therefore be used for different purposes, including labelling systems to help consumers to identify healthier food products, or as a standardised basis for the food industry to identify foods eligible to make health claims (Verhagen and van den Berg, 2008). It can also be used to provide a standardised and transparent basis for the application of food policies or regulations designed to improve diets and reduce population risk for NCDs.

Developing a robust nutrient profiling system requires clarity regarding the purpose of nutrient profiling, to enable the selection of appropriate index nutrients and reference amounts (Scarborough, 2007). In this section, we apply an evidence-based methodology to develop a nutrient profiling tool to improve diets and prevent NCDs in Fiji, using the systematic approach recommended by Rayner and Scarborough (Rayner et al, 2004), and then used by the World Health Organization to develop models in the European (WHO, Nutrient profile model, 2015) and Western Pacific Regions (WHO, Nutrient profile model, 2016 ). A detailed description of the methodology is provided in Appendix 1.

The purpose of this model is to provide clear criteria to identify food and beverage products which contribute to NCD risk (those high in fat, salt and sugar), which can be used as a basis to develop more targeted policy measures to reduce unhealthy consumption.

This model was developed using evidence-informed decision-making across the following 5 steps (Rayner et al, 2004): A. Selection of appropriate index nutrients, in relation to desired health outcomes; B. Consideration of the most appropriate choice of base (e.g. per 100g, per 100kJ); C. Identification of most appropriate type of model, in relation to intended use of model; D. Identification of the most appropriate and relevant nutrient thresholds, in relation to desired health outcomes; E. Validation and testing of the model against an objective measure of a healthy diet (in this case, food-based dietary recommendations for Fiji and the Pacific region), and in relation to representation of ‘healthy’ and ‘less healthy’ foods.

As the first step in this process, we identified food categories for the model. (Table 1). These were based on the WHO nutrient profiling model and informed by Pacific dietary patterns (outlined in the Background and Context).

Table 1: Detailed food categories

Cat 1: Choc, sweets, sugars Cat 1a: Choc, confect, toppings Cat 1b: Raw sugar, syrups

5 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

Cat 2: Cakes, sweet biscuits, sweet bakery goods Cat 3: Savoury snacks Cat 4: Beverages Cat 4a: Juices Cat 4b: Milk Cat 4c: Coffee, tea Cat 4d: Other beverages, sugar-sweetened beverages, drink mixes Cat 4e: Alcoholic beverages Cat 5: Edible ices Cat 6: Breakfast cereals Cat 7: Other dairy, yoghurt, cream, sour Cat 8: Coconut Products Cat 9: Ready-made, composite, convenience Cat 9a: Composite meals not commercial Cat 9b: Composite processed ready-made Cat 10: Butter, fats, oils (including coconut oil) Cat 11: Staples Cat 11a: Bread, crisp bread, roti, flour Cat 11b: Rice, grains, pasta Cat 11c: Noodles Cat 11d: Root crops and starches Cat 12: Fresh and frozen meat, poultry, fish Cat 12a: Fish, seafood Cat 12b: Poultry Cat 12c: Other meat Cat 12d: Eggs Cat 13: Processed meat, fish, poultry Cat 13a: Processed fish Cat 13b: Processed meat, poultry Cat 14: Fruit, fresh, frozen, dried Cat 15: Vegetables, fresh, frozen, Cat 16: Soya, tofu, legumes, nuts, seeds Cat 17: Sauces, dips, dressing Cat 18: Salt

A: Selection of appropriate index nutrients

Identifying foods and nutrients associated with NCD risk

Specific foods associated with increased risk of NCDs include those high in fat (particularly trans-fats and saturated fats), sugar and salt, while diets high in fibre, fruit and vegetables, fish provide a protective effect.

Dietary fats have a significant influence on NCD risk, particularly cardiovascular disease, due to their impact on blood lipids, blood pressure, thrombosis and inflammation (Katan, 2000). Cardiovascular disease risk reduces when saturated fats, found mainly in fatty meat and

6 Part 2: Nutrient profiling model processed foods, were replaced with polyunsaturated fats (de Souza et al, 2015). Processed meats, however, have an independent risk on NCDs, particularly colorectal cancer, coronary heart disease and diabetes (Micha et al, 2010).

Dietary salt intake plays a major role in influencing blood pressure and overall cardiovascular health (Sacks et al, 2001). Dietary salt intake is largely attributable to processed meats, packaged meal or snack foods, cooking sauces, bread and fast-food. A reduced sodium intake is known to decrease risk of stroke and fatal heart attack (He and MacGregor, 2013).

Dietary factors are also strongly linked to overweight and obesity. High BMI is an established independent risk factor for non-communicable diseases (NCDs) such as type 2 diabetes, cardiovascular disease and many cancers (Swinburn, 2011). Behaviours including snacking and eating frequency, eating foods outside of the home and eating refined sugars and carbohydrates increase obesity risk (FAO and WHO, Diet, nutrition and the prevention of chronic diseases).

Diets high in refined carbohydrates and sugar reduce the quality of the diet, increase risk of metabolic risk factors for NCDs and increase body mass because they are more likely to be energy dense and nutrient poor (Tordoff and Alleva, 1990).

A higher consumption of fruit and vegetables (particularly leafy green vegetables) is significantly associated with a lower risk of all-cause mortality, cardiovascular disease and Type 2 diabetes (Wang et al, 2014). Fruit and vegetables are a key source of dietary fibre, and an increased intake of fibre has been shown to reduce cardiovascular disease risk (Threapleton et al, 2013).

In Pacific Island Countries the key dietary pattern associated with increased NCD risk is what is termed the ‘modern’ diet (Huhes, 2003). The modern dietary pattern includes a high amount of processed foods (for example, instant noodles, processed meats, confectionary, packaged snack foods) and non-traditional foods (for example, cheese and eggs), and a reduced intake of local foods (for example, root crops), fruits and vegetables. A ‘modern’ dietary pattern is consistently associated with poorer nutrient intake and dietary adequacy, a worsened metabolic profile, higher body mass index and increased risk of NCDs (Bindon and Baker, 1985).

Given this, a nutrient profiling model relevant to diet-related NCD prevention in Pacific Island Countries like Fiji should address energy, sugar, sodium, fat, saturated fat and trans fatty acid, while promoting consumption of fish, fruit and vegetables, and nuts and legumes.

7 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

Consumption of foods and nutrients associated with NCD risk in Fiji

Analysis of data from Fiji’s 2014 National Nutrition Survey shows that Fijians aged 12-45 years consume an average of 2092Kcals per day, including 79.2g of protein (15.1 percent energy) and 61g of fat (26.2 percent energy), 302g carbohydrate (54.1 percent energy) and 23g of fibre (Table 2). They consume and average of 2178mg of sodium per day and 192g cholesterol. ‘Sugary foods’ contribute close to 15 percent to average carbohydrate intake (calculated based on consumption of foods ‘high in sugar’, due to absence of data — see Table 5).

A comparison of nutrient intakes to recommended values indicate that Fijian adults consume insufficient fibre, and adequate amounts of energy, salt and fat. However, given the high prevalence of overweight and obesity, and the fact that self-reported food intake data generally underestimates energy intake, caloric intake is likely to be much higher (see ‘data sources’ in Appendix 1).

Table 2: Average daily nutrient intake for Fijians over 12 years of age, compared to recommendations

Nutrient Average Intake Recommended dietary intake Levels Comparison for Fijians aged 12—45 yearsc Energy (Kcal) 2092 USDA: Males aged 19—51 years 2400 kcal For adult males and females, (active)f average energy intake is NRV ranges from 1238 kcal to 4,428 kcal a,e adequate but not excessive. - Girls aged 12 minimally Those requiring less energy that active=1,238Kcal/day consumed on average include - Girls aged 12 very active= 2,700Kcal/day younger (12—14 years) and - Males 30 minimally active= 2,404Kcal/day older women (51—70years) who - Males aged 30 very active=4,428Kcal/day are inactive or relatively inactive. - Women aged 60 minimally active= Average energy intake is around 1,547Kcal/day 60% higher than required young - Women aged 60 very active= women and older women who 2547Kcal/day are inactive. Protein (g) 79.2g Estimated Average Requirement:a Average protein intake is 52—65/day (male) adequate for the biological 37—46g/day (female) requirements of all population 35g/day (female 14—18 years) sub-groups. Recommended Dietary Intake: a 64—81g/day (male) 46—57g/day (female) 45g/day (female 14—18 years)

Fat (g) 60.9g <30% of energy b Contributes around 26% to total energy intake Saturated Fat Around 23.1% of <10% total energy intake b 23 % of total fat intake is 14g or (g) fat intake is from 126Kcal, well under 10% of total foods high in energy intake saturated fat Carbohydrate 301.9g Less than 10% of total energy intake from 45g (170Kcal) of carbohydrates % >46g of free sugarsb are consumed in sweet foods - Sugar ( %) carbohydrates (Table 4), which is less than 10 % from foods very total energy intake (equal to 54g high in sugarc carbohydrate)

8 Part 2: Nutrient profiling model

9 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

[Table 2 Cont’d] Nutrient Average Intake Recommended dietary intake Levels Comparison for Fijians aged 12—45 yearsc Fibre (g) 23.0g Girls 14—18 years= 22g/day 40 % less than recommended for Boys 14—15 years= 28g/day adults 51—70years Men 19—70 years=30g/day Women 19—70=25g/day

Sodium (mg) 2178mg SDT 5ga Consumption meets 3,992mgd recommended upper limits 5g salt/2 000mg sodium per dayb Iron (mg) 14.2mg Estimated Average Requirement: a Meets recommended 6mg/day (male) requirements — but small scale 5—8mg/day (female) surveys and high prevalence of Recommended Dietary Intake: a Iron Deficiency Anaemia suggest 8mg/day (male) significant sub-populations with 5—18mg/day (female) very insufficient intakes.

Zinc 8.8mg Estimated Average Requirement: a Meets recommended 12mg/day (male) requirements for women but is 6.5mg/day (female) inadequate for men Recommended Dietary Intake: a 14mg/day (male) 8mg/day (female) Calcium 548mg Estimated Average Requirement: a Very low average intake 840—1 100/day (male and female) Recommended Dietary Intake: a 1,100—1 300/day (male and female) Vitamin A (ug) 829mg Estimated Average Requirement: a Meets recommended 625/day (male) requirements 500/day (female) Recommended Dietary Intake: a 900/day (male) 700/day (female) Vitamin C (mg) 179mg Estimated Average Requirement: a Exceeds requirements 30mg/day (male and female adults) Recommended Dietary Intake: a 45mg/day (male and female adults) a Nutrient Reference Values (NRVs), National Health and Medical Research Council of Australia b Guidelines from the World Health Organization c Fiji’s NNS 2014 (as yet unpublished), authors own analysis d Fiji Salt Intake Assessment (FSIA) project (personal communication from The George Institute for Global Health) e Estimated energy requirements from NRVs factor in Basal Metabolic Rate (from age, size, gender) and Physical Activity Level f Dietary Guidelines for Americans, USDA, 2010

Throughout this analysis we highlight the contribution of discretionary foods to nutrient intakes. Australia's National Health and Medical Research Council describes discretionary foods as “foods and drinks not necessary to provide the nutrients the body needs, but that may add variety. However, many of these are high in saturated fats, sugars, salt and/or alcohol, and are therefore described as energy dense”. Similarly, Fiji’s Food and Health Guidelines call on Fijians to choose foods and drinks with less salt, sugar, fat and oil and

10 Part 2: Nutrient profiling model higher fruit and vegetable content. Discretionary foods can be consumed in small amounts, however they provide very few beneficial nutrients and are not necessary for human health. When these are consumed in greater amounts, they displace the opportunity to consume beneficial nutrients with excessive amounts of energy, fat, sugar and salt.

Energy and sugar intake Based on recent data from the 2014 National Nutrition Survey, staple foods provide the largest contribution to the energy intake of Fijians (45 percent), made up of root crops and starches (17.8 percent), bread, roti and flour (12.2 percent), and rice and semolina (11.5 percent). Composite meals, including home-made and ready-made account for 12.1 percent of energy intake by Fijians, followed by savoury snacks (9.0 percent). Fresh and frozen meat, fish and poultry (7.4 percent), beverages (4.9 percent) and sweet baked items including cake and sweet biscuits (4.2 percent) are also significant sources of energy (Table 3).

Table 3: Top contributors to intakes of calories, fat and sodium, by detailed food group (percentage contribution to total intake)

Calories Fat Sodium Staples (45.2%) Composite meals (all) (25.9%) Composite meals (not - Composite not commercially commercially processed) processed (23.5%) (19.0%) Ready-made, composite (12.1%) Staples (14.9%) Bread, crisp bread, roti, flour - Noodles (5.7%) (18.5%) Savoury snacks (9.0%) Fresh and frozen meat, poultry, fish Savoury snacks (16.9%) (10.5%) Fresh and frozen meat, poultry, fish Savoury snacks (10.3%) Added salt (13.9%) (7.4%) Beverages (4.9%) Butter, fats, oils (10.0%) Fresh fish, seafood (4.6%) - 2.9% sugar-sweetened - Vegetable oil (6.0%) Cakes, sweet bisc, sweet baked Coconut Products (8.8%) Processed fish (3.7%) (4.2%) Coconut Products (2.6%) Processed meat, fish, poultry (6.3%) Root crops and starches (2.7%) Butter, fats, oils (inc coconut oil) Cakes, sweet bisc, sweet baked Composite (processed ready- (2.6%) (6.1%) made) (2.6%) Sugar, confectionary, sweets (2.5%) Beverages (3.4%) Processed meat, poultry (2.6%) - 1.6% from raw sugar and syrups Processed meat, fish, poultry Noodles (2.3%) (2.5%) Soya, tofu, legumes, nuts, seeds Sauces, dips, dressing (2.3%) (2.5%) Source: 2014 National Nutrition Survey, National Food and Nutrition Centre, Fiji

Individual foods contributing the most to overall energy intake include cassava (13.6 percent), white rice (11.2 percent), roti bread (6.1 percent) and cabin biscuits (5.4 percent). Bread and flour together contribute to 4.8 percent of energy intake, followed by noodles chow mein (2.0 percent).

11 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

The main contributors to sugar intake are sugar-sweetened beverages (4.3 percent carbohydrate intake), cakes, sweet biscuits and breads (3.9 percent), plain sugar (2.8 percent), and chocolate and confectionary (1.4 percent).

Discretionary foods contributing significantly to energy intake in Fiji include savoury snacks (8.8 percent), cakes and sweet biscuits (4.2 percent), noodles (3.3 percent), sugar and confectionary (2.5 percent), sugar-sweetened beverages (2.9 percent), butters, fats and oils (2.6 percent) and highly-processed composite foods (1.6 percent). Together discretionary foods contribute 26 percent to total energy intake for the average Fijian.

Fat and saturated fat intake The category of food contributing the most to fat intake by Fijians is composite foods 1 (home prepared or commercial) (25.9 percent of fat intake). Other food categories contributing significantly to fat intake by Fijians includes staple foods (14.9 percent, with 5.7 percent contributed by noodles), fresh and frozen meat, poultry and fish (10.5 percent), savoury snacks (10.3 percent), butter, added fat and oils (10.0 percent, with 6.0 percent from polyunsaturated vegetable oil), coconut creams and products (8.8 percent). Processed meat (6.3 percent) and sweet baked items including biscuits and breads (6.1 percent) also contribute to fat intake by Fijians (Table 3).

Individual foods contributing significantly to fat consumption include coconut cream and products (7.5 percent), vegetable oil (6.0 percent), cooked fish with cabbage (5.5 percent), tuna in oil (4.3 percent), chow mein noodles (4.1 percent), cabin biscuits (3.6 percent), boneless chicken curry (3.5 percent) and roti (3.5 percent).

It is not possible to ascertain saturated fat intake through this dataset, however additional analysis shows that 23.1 percent of fat intake is from foods high in saturated fat (Table 6), including coconut cream, meat and commercial biscuits. Many composite meals (especially those commercially processed) would likely be prepared with palm-based vegetable oils, which are very high in saturated fat.

Coconut cream is particularly high in saturated fat (around 17 percent), however it plays a significant role in Pacific Island culture, cooking and eating, and is thought by many to be an important and protective food for Pacific Islanders (SPC, Leaflet No. 16). It is likely that the switch from a diet high in fat from coconut cream to a diet high in fat sourced from animal fat and processed foods is a key factor in cardiovascular disease development in the Pacific.

Around 18.8 percent of fat is being consumed through discretionary foods, including savoury snacks, sweet baked items and commercially processed composite foods. Foods to target for reductions in fat consumption include discretionary items being consumed excessively, and those foods and drinks, which are proportionally high in saturated fat (coconut cream, butter, ghee, palm-based oils, commercial biscuits and snacks, and commercial composite meals).

1 cooked or processed foods containing many different ingredients e.g. instant curries, soup mixes or meat pies. Nutrient composition of composite foods varies significantly between food items based on ingredients and the way each item is prepared. Most often composite meals are prepared with added fats and sodium.

12 Part 2: Nutrient profiling model

Sodium intake Around 70 percent of total sodium intake by Fijians is derived from four main food categories: composite meals (21.6 percent), bread and roti (18.5 percent), savoury snacks (16.9 percent) and added salt (14 percent). Other significant food categories include fresh seafood (5 percent), processed fish (4 percent), processed meat (2.6 percent), noodles (2.3 percent) and sauces and condiments (2.3 percent).

Intakes of beneficial nutrients It is important to ensure that interventions to reduce consumption of foods high in fat, salt and sugar do not inadvertently exacerbate deficiencies of nutrients important to growth, development and human function. Preliminary results from the 2014 National Nutrition Survey (unpublished) show that there are subsections of the Fijian population who remain at risk of micronutrient deficiency, particularly iron, vitamin A, folate and zinc (National Food and Nutrition Centre, NNS Data).

Our analysis shows that the food category ‘vegetables’ is an important contributor to Vitamin A consumption by Fijians, attributable for over 40 percent of intake. Other significant sources include composite foods (likely to be made with vegetables and meats) (25.1 percent), fresh and frozen meats (9.8 percent) beverages (5.5 percent), staples (5.0 percent) and fruits (5.0 percent) (Table 6). Food groups contributing most to fibre consumption include staple foods (40.1 percent), vegetables and nuts, legumes and seeds (14.8 percent), composite foods (13.3 percent) and savoury snacks (8.8 percent). For Vitamin C, top contributors are beverages (39.1 percent), staple foods (40.0 percent; 19 percent of which is from cassava), vegetables (11.4 percent) and fruits (10.4 percent) (Table 4).

Table 4: Top contributors to intakes of Vitamin A, Vitamin C and Fibre, by detailed food group (percentage contribution percent)

Vit A pp % Vit C pp % Fibre pp % Cat 15: Vegetables- fresh, frozen, 43.89 11.39 12.51 Cat 9: Composite foods- home and commercial 25.17 4.83 13.25 Cat 12: Fresh and frozen meat, poultry, fish 9.81 0.55 0.02 Cat 4: Beverages 5.48 39.06 1.84 Cat 11: Staples 4.95 30.96 40.15 Cat 14: Fruit, fresh, frozen, dried 4.85 10.41 3.89 Cat 10: Butter, fats, oils (inc coconut) 1.77 0.00 0.00 Cat 13: Processed meat, fish, poultry 1.21 0.04 0.02 Cat 3: Savoury snacks 0.39 0.76 8.79 Cat 16: Soya, tofu, legumes, nuts, seeds 0.25 0.26 14.75 Source: Authors’ analysis of the 2014 National Nutrition Survey

13 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

According to this analysis, protein consumption is on average adequate, with 30.0 percent sourced from fresh meat, fish and poultry, 23.0 percent from staple foods and 20.1 percent from composite foods. Average zinc consumption at 8.8mg meets recommended daily intakes for females, however it is short of what is recommended for the average males (12mg/day). Key contributors to zinc intakes are staple foods (36.9 percent), fresh meat fish and poultry (16.1 percent) and composite foods (16.1 percent).

Key sources of iron include staple foods (23.0 percent), pre-prepared foods (composite meals, such as curries) (21.3 percent) and fresh meat, fish and poultry (12 percent). Vegetables (10.9 percent) and savoury snacks (7.0 percent) also provide iron, the latter due to dried pea and lentil snacks. Fiji have a mandatory flour fortification program in place which increases the iron content of bread, roti and flour.

Surveys demonstrating high rates of anaemia among the Fijian population indicate that many people do not consume enough iron-rich food, or that foods contributing iron have low absorption availability (29). Haem iron is derived from animal proteins and has approximately 25 percent greater absorption rate than non-haem iron (48). Haem iron is found in just 14.5 percent of iron-contributing foods here, with the remainder made up with non-haem iron (Table 8). Non-haem iron is found in plant-based foods including lentils, vegetables and cereals.

In Fiji, which foods contain nutrients linked to diet-related NCDs in significant amounts?

An analysis of the packaged and pre-prepared foods available in Fiji showed that they contain significantly higher amounts of energy, fat, saturated fat, sugar and sodium when compared with unprocessed, whole foods (Table 5).

The most energy-dense food categories (per 100g) in Fiji include fats and oils, sweet toppings and confectionary, and sweet baked items including cakes, sweet biscuits and sweetened bakery items (eg donuts, buns, pancakes). Savoury snacks including biscuits and crisps, savoury bread items (including roti and bread), breakfast cereals, and noodles are also among foods with the highest energy density in Fiji.

The most sugar-dense foods available in Fiji included raw sugars and syrups, confectionary and toppings, cakes and other sweet baked items. Breakfast cereals on average contained more sugar than edible ices and sugar-sweetened drinks. Sauces, dips and dressings also contained significant amounts of sugar.

The foods with the highest fat density in Fiji included fats and oils, chocolate and confectionary, coconut products, savoury snacks, meat, processed meat and poultry. Fats and oils, chocolate and confectionary, other dairy items (e.g. cheese, cream), noodles, processed meat and savoury snacks contained the highest proportion of saturated fatty acids.

14 Part 2: Nutrient profiling model

Apart from table salt, foods containing the highest amounts of sodium per 100g includes: noodles; prepared sauces, dips and dressings; savoury snacks; composite and convenience foods; and processed meat and poultry. Bread and roti also contain high levels of sodium, and this category is a key contributor to sodium because of the quantities in which it is consumed by Fijians.

Some specific examples of processed foods containing high levels of ‘nutrients of concern’ are listed below: - Savoury snacks contain extremely high levels of sodium (1191mg) and an average of 20 percent fat with 8 percent saturated fat. - Packaged noodles contains 1320mg of sodium per 100g and 13.8 percent fat, over half of which is saturated (7.2 percent). - Bread and flour-based products (bread, roti) contain around 760mg sodium per 100g, and contribute over 18 percent of total sodium intake. - Processed meats are comprised of an average of 17.7 percent fat and 8.3 percent saturated fat. They contain excessive levels of sodium (average 902mg/100g). - Sugary drinks including soda, tang and juice contain between 7 percent and 13 percent sugar. When consumed as a 325ml serve, a drink with 10 percent sugar equates to 32g of sugar (more than 6tsp of sugar). - Sauces, dips and dressings are especially high in sodium (1266mg/100g) and contain significant amount of sugar (average 15 percent content). - Confectionary, toppings and chocolate are especially high in sugar (average 56 percent) but provide little or no quantities of beneficial nutrients (e.g. protein, iron, vitamins). - Composite and cooked meals are highly variable in their nutrient content due to the potential mix of ingredients and cooking methods. Composite meal examined here are on average high in energy (avg>400Kcal per 100g) and very high in fat (avg 25 percent fat). They are often prepared with high levels of salt, wtih composite foods processed commercially (category 9b) delivering on average around 1400mg of sodium per 100g serve. - Cakes, sweet biscuits and sweet baked items contain around 419Kcal per 100g serve, comprised of around 30 percent sugar 15 percent. These foods account for 5 percent of total energy intake by Fijian, but deliver very few other beneficial nutrients to Fijians (ie iron, vitamin C, vitamin A). - The range of breakfast cereals available in Fiji are especially high in sugar, with the average cereal containing 23 percent sugar. Malt-o-Meal cereals, Fruity Dyho Bites and Tootie Fruity contain between 40 percent and 46 percent sugar. Kellogg’s Frosted Flakes and Cocopops each contain 36 percent sugar and Kellogg’s Smores and Frosties contain over 40 percent sugar. - Fats containing especially high ratios of saturated fat includes butter (49 percent saturated), ghee (60 percent) and coconut oil (82 percent), are each consumed in large amounts in Fiji. ‘Vegetable’ fats and oils imported from Malaysia and Indonesia and generically marked ‘vegetable oils’ are likely palm-oil based. Palm oils contain between 40 and 46 percent saturated fat (a vegetable fat that is very high in saturated fat)(12). - Some packaged foods deliver a higher number of beneficial nutrients and relatively low levels of sugar, salt and fat, for example split peas or unsalted nuts.

15 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

16 Part 2: Nutrient profiling model

Table 5: Energy, fat, sugar and sodium density of packaged and fresh foods in Fiji

Average nutrient density of foods Detailed Food Category Energy Fat (%) Sat fat Sugar ( %) Sodium Source (Kcal/100g) % (mg/100g) Cat 10: Butter, fats, oils (inc 653 73 28 0 486 PSS coconut) Cat 9b: Convenience, ready- 485 26 n/a n/a 1404 NNC/PFCT made Cat 9a: Composite meals 442 24 n/a n/a 751 NNC/PFCT Cat 1a: Choc, confect, 423 29 17 56 168 PSS toppings Cat 2: Cakes, sweet bisc, 420 15 7 30 317 PSS sweet baked Cat 3: Savoury snacks 429 20 8 1 1192 PSS Cat 6: Breakfast cereals 378 4 1 23 378 PSS Cat 1b: Raw sugar, syrups 358 0 0 87 66 PFCT Cat 11a: Bread, crisp bread, 307 6 n/a n/a 754 PFCT/PSS roti Cat 11c: Noodles 312 13 7 2 1321 PSS Cat 12c: Other meat 279 20 n/a 0 118 PFCT Cat 7: Other dairy, yog, 250 21 16 9 598 PSS cream, sour Cat 8: Coconut Products 234 23 n/a n/a n/a PFCT Cat 12b: Poultry 234 15 n/a n/a 110 PFCT Cat 12d: Eggs 233 15 n/a n/a 146 PFCT

Cat 5: Edible ices 223 12 22 49 PSS Cat 13b: Processed meat, 183 18 8 1 903 PSS poultry Cat 12a: Fish, seafood 160 7 n/a n/a 211 PFTC Cat 13a: Processed fish 136 7 2 2 457 PSS Cat 17: Sauces, dips, 131 6 2 15 1266 PSS dressing Cat 11d: Root crops and 124 1 n/a n/a 28 PFCT starches Cat 11b: Rice, grains 123 0 0 0 5 PFCT Cat 16: Soya, tofu, legumes, 118 0 0 3 2 PFCT nuts, seeds Cat 14: Fruit, fresh, frozen, 93 1 n/a n/a 16 PSS & PFCT dried Cat 15: Vegetables, fresh, 61 2 1 4 181 PSS & PFCT frozen, Cat 4b: Milk 58 2 1 5 42 PSS Cat 4d: Other bev, ssb, choc 52 1 0 10 22 PSS powder, Cat 4a: Juices 42 2 0 9 42 PSS Cat 4c: Coffee, tea 1 1 0 0 0 PSS Cat 18: Salt 0 0 0 0 38 000 USDA PFCT: Pacific Food Composition Tables, with author analysis to create category averages [15] PSS: Pacific Store Surveys, with author analysis to create category averages [14] USDA: United States Department of Agriculture Food Composition Database [16]

17 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

Based on the nutrient analysis performed here, ‘discretionary foods’ in Fiji include: - Chocolate, confectionary and toppings, sugar and sugary syrups (e.g. honey) - Cakes, sweet biscuits and other sweet bakery items (e.g. iced buns) - Edible ices (e.g. ice creams and frozen ices) - Savoury snacks (e.g. extruded snacks, including noodles) - Sugar sweetened beverages (e.g. soft drinks, powder mixes and flavoured milks) - Instant noodles

In Fiji, the most nutrient-dense food categories include root crops and starches (energy, fibre and vitamins), fresh fish and seafood (protein), fruit and frozen fruit and vegetables (fibre, vitamins and minerals), lower fat dairy products (calcium) and fresh, leaner cuts of meat and poultry (protein, iron and zinc).

Summary: Foods and Nutrients to target in Fiji

This analysis suggests that policies aiming to reduce diet-related NCD risk should address excessive consumption of energy, sugar, sodium, fat, saturated fat and trans fatty acid.

These data indicate that policies targeting foods contributing significant salt, sugar and fat consumption to the diet would not significantly impact on intakes of beneficial nutrients protein, zinc, iron, calcium, Vitamin A or fibre.

Composite meals, bread and roti, and fresh meat and poultry, are all core foods which contribute significantly to intake of beneficial nutrients in addition to nutrients linked to NCDs, sodium, fat, saturated fat, therefore these groups require specific consideration as to how proposed thresholds are applied in an appropriate policy intervention.

Beneficial nutrients to be promoted by associated policies include calcium, iron, zinc and fibre.

Step B: Consideration of the most appropriate choice of base (e.g. per 100g, per 100kJ)

According to Fiji’s Food Safety Regulations (2009), all package foods and beverages must on their label indicate energy, protein (g), fat (g) and carbohydrate (g)content per 100g or 100ml.

The Pacific Store Survey found that nearly all food products provided data per 100g, consistent with food standards in Australia and New Zealand. The predominating source of manufacture for packaged foods in Fiji at that time was Australia (26 percent), Fiji (21 percent) and NZ (13 percent). A 2017 study confirmed that the most common origin of food products imported to Fiji included Australia, NZ, Japan and US(17) (Ravuvu et al, 2017).

Therefore, a base of per 100g/per 100ml is selected as the appropriate base.

18 Part 2: Nutrient profiling model

Step C: Identification of most appropriate type of model, in relation to intended use of model The purpose of this nutrient profiling model is to underpin policy(ies) that aim to incentivise healthy food consumption (and/or disincentivise unhealthy food consumption). Therefore, the primary considerations for Fiji are that the model: - Is straightforward to implement with respect to policy administration; - Clearly identifies healthier and less-healthy foods to which the policy incentives can be applied, in a transparent and systematic manner; and - Improves the food environment (availability, affordability etc) through incentivising production and/or import of healthier products

Based on this, the most appropriate model would be category specific (provides incentives for consumers AND producers/importers to switch to healthier products within categories; policies can also be applied to entire categories to incentivise movement from less healthy categories to healthier categories) and threshold-based (provides straightforward nutrient criteria for policy administration, that incentivise consumption, production and/or import of healthier foods). This is a similar approach taken by the World Health Organization in the Western Pacific and European Regions for a nutrient profiling model to underpin restrictions on marketing of foods and beverages to children.

Step D: Identification of the most appropriate and relevant nutrient thresholds

Goals established by the consumption analysis (Step A) include to: - Reduce energy and sugar intakes ‘discretionary foods’, which are energy dense and/or nutrient poor - Reduce consumption of saturated fat through discretionary foods, processed meat, fresh meat and poultry and composite foods (home and commercially prepared). - Reduce fat intake from sources high in saturated fat and trans-fatty acids - Reduce sodium use in prepared foods and manufacture of bread and processed foods - Promote consumption of fresh fish, fruit, vegetables and legumes - Increase consumption of foods high in fibre, calcium, zinc and iron

Groups to be exempted from thresholds in the model Some groups have not been included in the model because they are either core foods providing large amounts of beneficial nutrients, or they are groups shown to be insignificant to Fijians. Reducing the number of categories also reduces administrative burden.

Fruit, vegetables, starchy vegetables: These are core foods, and processed varieties are being consumed in minimal amounts (fried, salted varieties are included under savoury snacks) Coconut products: juice and milk and cream

19 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

Table 6: Mean nutrient composition of foods and categories being targeted by Fiji’s nutrient profiling model

Detailed food categories used during Composition Composition Composition Composition Composition analysis. Mean Mean Fat ( Sat fat % Mean Sugar Mean energy %) ( %) Sodium (Kcal/100g) (mg/100g) Choc, confect, toppings syrups 423 29 17 56 168 Raw sugar, 358 0 0 87 66 Cakes, sweet bisc, sweet baked 420 15 7 30 317 Breakfast cereals 379 4 1 23 420 Savoury snacks 429 20 8 1 1192 Sweetened beverages 52 1 0 10 22 Edible ices 223 12 8 22 49 Composite meals 174 5 2 4 1061 Convenience, ready-made 174 5 2 4 1061 Butter, fats, oils (inc coconut) 653 73 28 0 486 Bread, crisp bread, roti, flour 307 6 n/a n/a 754 Noodles 312 14 7 2 1321 Fish, seafood 160 7 n/a n/a 211 Poultry 234 15 n/a n/a 110 Other meat 279 20 n/a 0 118 Processed fish 136 7 2 2 457 Processed meat, poultry 183 18 8 1 903 Sauces, dips, dressing 131 6 2 15 1266 Table salt 0 0 0 0 38 000

Note: Data sources for each category detailed in Table 5 above

20 Part 2: Nutrient profiling model

Proposed threshold groups and nutrient thresholds Based on food categories above, foods identified for targeting, and data on current composition of foods in Fiji, thresholds for fat, sugar and sodium are proposed for all food groups, with reference to international models (Table 7).

Table 7: Summary of considerations in developing nutrient profiling model for Fiji

Nutrients WHO WPRO Other reputable Proposed Notes (to be deleted) targeted in Model [18] models[19, 20] for Fiji MM= mean/median when they are proposed food similar values categories All thresholds per 100g Choc, confect, Nil given- not Danish Code <5g fat Mean fat: 29% median: 10% toppings, syrups permitted Marketing: <5g sugar Mean sugar: 56%, median sugar 56% Desserts and candy <5g fat, < 5g sugar Raw sugar Nil given- not Not applicable Not allowed in most models Table salt permitted

Beverages Juice 5g sugar Danish Code <5g fat Mean and median sugar 9-—10% (sweetened Milk 4g fat Marketing: <5g sugar drinks, juice, SSB 0g sugar Beverages 0g sugar, 5% to allow milk which is naturally 5% milk) 0g fat sugar Netherlands Tripartite: sugars 6g, saturated fat 0.5g Other dairy Yogurt, cream 4g Danish Code <5g sugar Sugar is a concern in yogurt, in all (excluded milk fat, 10g sugar, Marketing: <20g fat other dairy it is fat. and butter) 80mg sodium Yogurt, cream, curd Established sugar threshold to limit Cheese 20g fat, 2.5g fat, 10g sugar sugary yogurt and fat threshold to limit 520mg sodium Cheese, 20g fat overly fatty creams and cheese?

Edible ices 10g sugar Danish Code <5g fat Mean and median sugar:22%(twice the 4g fat Marketing: <8g sugar sugar of SSB), Mean and median fat 80mg sodium Desserts and candy 11—12%t 5g fat, 5g sugar Ices are generally consumed in small quantities. This threshold promotes 100% juice watered down (~9% sugar), while ruling out SSB Cakes, sweet Nil given- not HF: bread <400mg <400mg Mean fat: 15%, median fat 17$ biscuits permitted sodium sodium Mean and median sugar 30—31% Danish Code <10g fat Mean sodium: 316mg, median sodium Sweet baked 10g sugars Marketing: <10g fat <10g 265 (but most from biscuits). buns, breads <10g sugar sugar Large sugar and fat reduction for biscuits (but they discretionary). However, sodium in this group should match ‘bread’ as much of the energy consumed in this group was largely sweetened bread, buns and dumplings as opposed to commercial biscuits. Therefore increased to 450mg.

21 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

Table 7 [Cont’d] Nutrients WHO WPRO Other reputable Proposed Notes (to be deleted) targeted in Model [18] models[19, 20] for Fiji MM= mean/median when they are proposed food similar values categories Breakfast cereals 10g fat Danish Code <10g fat MM around 23 g sugar, 11g fat and 15g sugars Marketing: <15g 470sodium. 640 sodium 10g fat, 15g sugar sugar Integrating this group with the 2 above <450mg due to similarities. sodium Weetbix, cornflakes, oats, rice bubble all below 10g sugar. Of these, only weetbix and oats are also under 450mg sodium above. Around 50% of cereals <450mg sodium, however they are high in sugar. Demonstrates that 450 is feasible for cereal. Future proofing: use ‘added sugar’ as a number of low sodium mueslis would comply Bread, crisp 480mg sodium HF: <400mg sodium <400mg Punjas roti and standard white bread bread, roti 10g fat Danish Code sodium currently 750mg+, so 450mg 10g sugars Marketing: 10g fat, represents a 50% reduction. 10g sugar, <10g fat 400—450mg is consistent with other Finland better models. According to market research choices: sodium by HF (2010) there were plenty of <700gm for Heart breads contain around 400mg/100g symbol [21] [22]. Pacific Salt Targets Should align to Pacific Salt Targets 400mg sodium Savoury snacks 400mg sodium Netherlands <450mg Mean fat: 20%, median 26% 0 sugar Tripartite: sodium Mean sodium: 1191, median sodium <4g saturated fat 652mg Pacific Salt Targets allow 560—800mg depending on the food, however, there is an argument for remaining strict because this group is completely discretionary. Noodles 10g fat Danish Code <450mg Only 2 brands have <450mg, only 4 10g sugar Marketing: sodium brands have <10% fat. 480g sodium 500mg sodium <10g fat Strict because noodles are 10g fat discretionary (I believe) and overeaten 10g sugar

22 Part 2: Nutrient profiling model

Table 7 [Cont’d] Nutrients WHO WPRO Other reputable Proposed Notes (to be deleted) targeted in Model [18] models[19, 20] for Fiji MM= mean/median when they are proposed food similar values categories Composite meals 10g fat HF: ready meals <10g fat, Allows some oil added salt while (prepared, 4g saturated fat 2g saturated fat <4g discouraging excessive use. convenience and 10g sugar 300mg sodium saturated Sodium represents 50% reduction ready-made) 400mg sodium HF: canned meat <225 Kcal from mean (over 750) 225 Kcal meals and ready Consistent with other models meals> 75g veg per 500mg Promotes added vegetable content per serve sodium serve Danish Code Marketing: Minimum 500mg sodium, 10g 75g fat vegetable 10g sugar content B Danish gives special per serve conditions for of food composite Butter, fats, oils 20g saturated fat HF: <20% of total fat 20g Consistent with other models, many (inc coconut oil) 560mg sodium saturated, <1% trans saturated healthier alternatives available locally fat Mean 28g, Median 16g Netherlands Future proofing- should consider trans Tripartite: Spreads fats limits. and fats 16% Could include sodium but no data saturated provided in Pacific nutrient composition tables Fresh fish, <20g fat Danish Code 20g total Targeting reduction of fatty meat cuts seafood, poultry, Marketing: fat Proposed merge with group below other meat 20% fat Processed fish, <20g fat HF: canned meat 20g fat Aligned to Danish model- WPRO model meat, poultry <680mg sodium meals 4g is very lenient on this criteria 285Kcal saturated 500mg allows most canned fish 1.5g saturated fat fat (460mg/100g) and very few ultra- No trans fat 500mg processed meats) 350mg sodium sodium Danish Code Marketing: 500mg sodium Netherlands Tripartite: 4g saturated fat Sauces, dips, 400mg sodium Danish Code 400mg Mean fat is 6g, mean sodium 1600mg dressing 10g fat Marketing: 10g fat sodium Sauces are discretionary but can be Netherlands helpful to increase veg consumption Tripartite: 2g Only spices and herbs are under saturated fat 400mg, though most curry pastes and some mustards <500mg. Could potentially lift to 500mg to allow these 2. Fresh fruit and No threshold proposed Processed veg and fruit minimal- fruit vegetables salad in syrup minscule Coconut No threshold proposed Oil included in fats and oils products

23 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji b Danish code specifies for composite foods, all must contain 3 food groups in proportion with “the plate model”

Step E: Validation and final nutrient thresholds for Fiji

The model was validated with 3 nutrition experts working in Fiji.

Summary of validation The 18 food categories and 11 threshold groups were modified to 13 threshold groups for the following reasons:  ‘Other dairy’ was inserted as a threshold group given likely future market growth in Fiji.  One threshold group was split into 2 following feedback about sodium thresholds. o Bread was reduced to 400mg from 450mg to align to Pacific salt targets o Noodles and savoury snacks remain conjoined with shared sodium and fat thresholds

The model was improved for clarity regarding inclusion of pastries, tinned vegetables, honey, and powdered drink mixes.

It was agreed that the model had provided an appropriate indication of the healthfulness of foods.

24 Part 2: Nutrient profiling model

Final proposed Nutrient Profiling Model for Fiji The following model is thus proposed for use in Fiji (Table 8).

Table 8: Nutrient profiling model with thresholds proposed for Fiji

Threshold Food Category group To be considered ‘healthier’, product must contain less than: Fat g/ Saturated Sugar Sodium Additional criteria 100g Fat g/100g mg/100g g/100g

1 Choc, confect, honey, 5 5 toppings, syrups 2 Raw sugar Table salt Not applicable

3 Beverages (sweetened 5 5 In future, ‘added drinks, juice, milks — sugar’ and reduce including powdered to zero. mixes) 4 Other dairy (cheese, 20 5 yogurt, cream, canned) 5 Edible ices 5 8

6 Cakes, sweet biscuits, 10 10 400 pastries Sweet baked buns, breads Breakfast cereals In future, use ‘added sugar’ to distinguish 7 Bread, crisp bread, roti, 400 flour 8 Savoury snacks, nuts Noodles, rice, grains 10 450 9 Composite meals 10 4 10 500 >75g vegetable (prepared, convenience content per serve and ready-made) 10 Butter, fats, oils (inc 20 In future, consider coconut oil) adding trans-fatty acid limit 11 Fresh fish, seafood, 20 4 500 In future, consider poultry, other meat adding trans-fatty acid limit Processed fish, meat, poultry 12 Sauces, dips, condiments 400 13 Fresh and minimally processed fruit and vegetables Not applicable Coconut products (no other ingredients)

25 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

Part 3: Identify fiscal policy and priority complimentary measures for incentivizing access to affordable healthy foods in Fiji

The Rome declaration of the Second International Conference on Nutrition highlights the importance of using policy to create a healthier food environment, with improved availability, accessibility, affordability and acceptability of healthy foods relative to unhealthy foods (FAO and WHO, Rome Declaration on Nutrition). For example, policies that support public procurement of healthier food in schools and other settings; and marketing, labelling or fiscal policies; targets for reformulation to reduce salt, fat and sugar in processed foods; agricultural policies to increase production of fruits and vegetables; and policies to increase the amount of healthier food sold by retailers or street vendors.

In this section of the report, we summarise the current policy context in Fiji, assess the policy implications of the nutrient profiling model, and identify specific policy measures.

Review of policy context

The Government of Fiji has identified non-communicable diseases (NCDs) as a priority. Cabinet endorsed the NCD Strategic Plan in 2016, including creation of a multisectoral steering committee, which will be chaired by the Prime Minister. Nutrition is also a policy priority for the Government of Fiji, and it is planned that it will feature significantly in the new National Development Plan, currently under development.

Comprehensive action on nutrition and NCDs is described in the Food Security and Nutrition Policy for Fiji, a whole-of-government policy led jointly by the Ministries of Agriculture; Health and Medical Services; Education, Heritage and Arts; and Women, Children and Poverty Alleviation. The strategic goal of the Food Security and Nutrition Policy is to ensure the availability, accessibility and affordability of safe and nutritious food for every Fijian, sufficient to meet their dietary needs, cultural and food preferences for an active and lasting healthy life.

This policy includes ‘supply side’ focused policies and programmes to promote availability, accessibility, affordability and acceptability of healthier foods. For example, promotion of sustainable, diversified and resilient food systems, investment in nutrition-sensitive value chains, support for healthier school food environments, and investment in nutrition sensitive social welfare initiatives. The Policy is operationalized by the Food Security and Nutrition Action Plan, which identifies specific policy changes, initiatives and programs to achieve nutrition objectives.

The key policy document for the Ministry of Agriculture is the Strategic Development Plan 2018-2022. The Ministry of Agriculture is supporting the Zero Hunger Initiative, including through a multi-sectoral stakeholder committee.

26 Part 3: Priority policy options

Within the health sector, other relevant food and nutrition policy documents and governance structures include the Food safety legislation and associated Technical Advisory Group (Food TAG), National Breastfeeding Policy, the School Health Policy (2016), and the Adolescent Health Policy.

The Ministry of Health has also developed government catering guidelines, which are in preparation for submission to Cabinet, and regulations on the marketing of foods and beverages to children, which are currently with the Solicitor-General’s Office.

The Ministry of Education and Ministry of Health have formally collaborated on the School Health Policy (2016) and ongoing Health Promoting Schools Initiative. Relevant to this, the Ministry of Health, Women, Social Welfare and Poverty Monitoring is also supporting the initiative “Healthier Fiji schools” through assessing the opportunity to expand the free school milk programme with a fruit plate supplement.

Existing specific taxes on unhealthy foods include import excise taxes of 15 percent (in addition to fiscal import duties of 32 percent) on confectionary and chocolate, prepared foods, sweet biscuits, potatoes, and bottled soft drinks (Gov of Fiji, Customs Tariff Schedule). There are fiscal import duties (no excise taxes) on fresh fruit and vegetables of 5 percent, except for tomato, lettuce, cucumber, aubergine, pumpkin and potato, which have a fiscal import duty of 32 percent. It should be noted that this follows significant liberalization and reductions in tariffs since the 1960s (Thow et al, 2011).

Policy implications of the model

Sweets, confectionary, frozen ices: These contain high amounts of sugar and in some products fat. They are mostly ultra-processed and contain no beneficial nutrients, and are subject to a sugar threshold (thus capturing those varieties that may also be high in fat, as all are high in sugar). They are a discretionary food and an appropriate target for taxation.

Salt and Sugar: Salt and sugar in their free forms are not amenable to thresholds, and policy intervention should target consumer behavior. For example, the addition of salt at the table, or addition of sugar to hot beverages. Social marketing may be the most appropriate approach.

Beverages: Sugar sweetened beverages are a discretionary food, contributing to sugar intakes but not beneficial nutrient intakes. The literature suggests that consumption is highly amenable to reductions through taxation.

Sweet baked items: Sweet baked items including cakes, biscuits and buns are high in energy from fat, sugar and carbohydrates, thus subject to sugar and fat thresholds. They are discretionary foods and thus appropriate targets for taxation.

Breakfast cereals: Although these are not commonly consumed in large amounts by Fijians, global trends suggest consumption will slowly increase (Euromonitor International, Cultural Variations or Peculiar Tastes: Global Prospects for Breakfast Cereals). Breakfast cereals in Fiji are

27 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji extremely high sodium and sugar content. At the moment, they can be considered a discretionary food, and are an appropriate target of taxation; this may also help to limit market growth in unhealthy cereals.

Bread and roti: These are core foods in Fiji, but they also contribute significantly to sodium intake because of the high rate of consumption. As core foods, policies targeting sodium consumption in bread should focus on reducing sodium content – in particular, incentivizing reformulation – as opposed to reducing quantities consumed, due to their significant contribution to protein, calcium, zinc, fibre, Vitamin A and iron intake.

Savoury snacks: Savoury snacks are highly processed and contain high levels of fat, saturated fat and sodium, with each shown to be a significant source of intake for Fijians. As discretionary foods, they are likely to be amenable to taxation and also to incentives for reformulation.

Noodles: Packaged noodles are contributing significantly to intakes of energy, fat and sodium, all of which will be targeted with nutrient thresholds.

Composite meals (home prepared and commercial): Composite meals have been shown to be significant contributors to most beneficial nutrients (protein, iron, calcium, zinc, vitamin A and fibre). However, these foods are also a major contributor to intakes of energy, sodium and fat in Fiji. Composite meals (non-commercial) may be difficult to target with tax or other measures based on composition, because they are prepared or sold without food labels and therefore nutrient composition data. Composite meals are, however, amenable to reformulation, such as through increasing vegetable content and reducing salty flavorings or saturated fat content, and policies should target this approach.

Fats and oils: The contribution of fats and oils to NCD-risk is highly variable, dependant on the product’s source and level of saturation or hydrogenation. Butter, ghee and low-quality vegetable oils (i.e. palm oil) can be a major contributor to intake of saturated and trans fatty acids, and therefore subject to a saturated fat threshold.

Fresh meat, fish and poultry: Most fresh meat and poultry is sold free of packaging and nutrient labelling. However, these foods contribute 10 percent to overall fat intake by Fijians and many are high in saturated fat. Policies should promote reduction in consumption of the higher fat cuts of meat (for example chicken wings) – these may be an appropriate target for social marketing. For example, the social marketing that accompanied the sales ban on mutton flaps helped to shift population perceptions and demand for high fat sheep meat (Thow et al, 2010).

Processed meat and fish: Processed meat and processed fish contribute relatively equally to intakes of sodium, fat, and beneficial nutrients. However, both are high in sodium, and processed meat in particular contains fat which is largely saturated. The thresholds for fat and sodium are likely to be a strong basis for incentivizing reformulation, for example, through application of nutrient targets or mandatory nutrient limits.

28 Part 3: Priority policy options

Recommendations for fiscal policy and complementary measures

Fiscal policy measures

This analysis suggests the application of significant specific excise taxes on discretionary foods not meeting nutrient profiling criteria, as these foods contribute significantly to intakes of unhealthy nutrients and are likely to be more price sensitive than core foods.

Discretionary food categories contribute an average of 149g of food to the daily intake (9.2 percent of 1615 grams/person/day). However, these products contribute a much higher proportion of nutrients of concern: 343 calories (16.4 percent of 2091 calories/person/day), 15.4 grams of fat (25.2 percent of 61grams/person/day) and 440 milligrams of salt (20.2 percent of 2176 milligrams/person/day). As a result, reductions in consumption of these products can have a substantial effect on consumptions of nutrients associated with NCD risk. Many of these foods are relatively inexpensive for the significant contribution they make to intakes of energy and nutrients of concern, and thus appropriate targets of taxation (as the cost does not internalize their significant health impacts).

The literature reviewed indicates that significant taxes, of up to 50 percent or 100 percent, on these products would have a meaningful impact on consumption and health. The specific food categories (discretionary foods) recommended to be taxed are:  Confectionary and sugar  Beverages (sweetened drinks, juice, milks)  Edible ices  Cakes, sweet bakery and biscuits  Savoury snacks, including instant noodles

In addition to these considerations of impact and price elasticity, the recommendation to target only discretionary foods is also influenced by considerations of feasibility. The application of fiscal policy measures to all food categories in the nutrient profiling model is limited somewhat by the requirements for labeling in Fiji. Fiji’s Food Safety Regulations (2009, amended 2012) require only that food labels specify energy, protein, fat and carbohydrate content, and not sugar or saturated fat.

In Part 4, we model the estimated impact of a tax on discretionary foods at 20 percent, 50 percent and 100 percent on consumption, health, revenue and household expenditure.

Complementary policy measures targeting discretionary food consumption There is opportunity to utilize part of the revenue generated by taxation through complementary measures targeting the taxed foods. As described in the review in Part 1, this would act to increase the effect of taxation on diets and health.

1. Additional incentives for reformulation We recommend establishing nutrient targets for salt, fat and sugar in processed foods, based on the nutrient profiling. This would compound the effect of the tax in encouraging

29 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji reformulation, and also improve the nutrient composition of core foods and composite foods, which are not affected by the tax.

2. Complementary social marketing A social marketing and school-based campaign that de-normalises consumption of discretionary foods would enhance the effect of the tax, promote public awareness and support for the tax, and further promote reductions in discretionary salt and sugar consumption.

3. Restrictions on marketing of discretionary foods There is an opportunity to enhance the draft regulation on marketing of foods and beverages to children with a broader regulation that limits the marketing of discretionary foods that do not meet the nutrient profiling criteria. This would complement the social marketing campaign and further incentivize reformulation, through targeting public acceptability of discretionary foods.

Complementary policy measures to promote healthy foods

The significant revenue that would likely be raised by the tax would provide financial support for implementation of measures recommended in the Food Security and Nutrition Policy and a Food Security and Nutrition Action Plan that target healthy food affordability and availability. This would further support consumers to make healthier choices, through making these more acceptable and accessible. In particular, subsidies to increase affordability of staple root crops, wholegrain foods, fruit, vegetables and legumes; and subsidies/technical support to increase use of healthier inputs (healthy oils) and reductions in unhealthy components in processed food manufacturing.

30 Part 4: Modeling of fiscal policy impact

Part 4: Model the impact of a science-based health excise on consumption, nutrition and revenue

Price elasticity estimates, based on international evidence Price elasticity of demand refers to changes in consumer demand, in response to a change in price of a good. Price elasticity estimates (PEEs) are almost always negative, because demand falls as price rises. Food items that can be considered necessities such as staples such as bread and rice tend to be price inelastic, meaning that it has a price elasticity of demand between zero and one (demand falls less than one percent, for a one percent increase in price), and may be in the range of 0 to -0.4, for example (Andreyeva et al, 2010). In contrast, discretionary foods such as chocolate and confectionary may be more price elastic in their demand; for example, around -1. Specific foods may have higher price elasticities of demand, due also to the ability of consumers to substitute between such foods. It is this substitution that provides the mechanism by which fiscal measures (taxes and subsidies) can be employed to encourage healthy diets. Examples of such substitution might be low-fat meats for high-fat meats, or unsweetened beverages for sweetened beverages.

Overall, price elasticity values tend to be higher in low and middle-income countries compared to high-income countries, as consumers have a higher level of price sensitivity (Green et al, 2013). However, as indicated earlier, staple foods, which vary according to cultural factors, are usually very inelastic. In Table 1 we provide estimates for Fiji.

International research suggests that price elasticities alone do not account for consumer reactions to large health-related taxes, because these are likely to be fortified by complementary consumer awareness and education interventions (Thow et al, 2014). This highlights the potential for the modelling here to underestimate the impact of fiscal policy that is implemented as part of a package of interventions, as proposed in Part 3. While recent research on the use of PEEs to estimate the impact of taxation of soft drinks in the Pacific suggests that these might overestimate consumption effects (Gibson and Romeo, 2017), the cumulative effect of multiple interventions is unaccounted for. As such, these estimates should be reasonably robust for targeted fiscal policies implemented as part of a policy package designed to disincentivise consumption of discretionary foods.

Price elasticity values for discretionary foods in Fiji Based on a review of the international literature, and with particular emphasis on price elasticity estimates from low- and middle-countries (LMIC), we have identified PEEs for the specific food categories in the nutrient profile model relevant to Fiji (Cornelsen et al, 2013).

Sweetened beverage price elasticities, relevant to LMIC, have been identified in a number of reviews. Snowdon estimated a PE of between 0.7 and 12 for soft drinks in both Fiji and Tonga. This is consistent with recent analyses from Brazil and India, which generated PEEs of 0.85 and 0.94 for sugar-sweetened beverages, respectively (Basu et al, 2012). It is also

2 Note that as per convention we present PEEs in absolute values – but all of the PEEs presented represent negative PEEs (demand falls as price rises)

31 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji consistent with a more general price elasticity estimate for “drinks and sweets” of 0.96 in low-income countries and 0.61 in middle income countries, in Muhammad (2011). However, a more recent analysis from Ecuador found an own-PEE of 1.2 for the total population, and 1.3 for the low socioeconomic group (Paraje, 2016).

We therefore propose a PEE of 1 for sweetened beverages in Fiji. There is also potential for complementary interventions, such as public awareness campaigns regarding health risks and consumer-oriented labelling, to increase this price elasticity.

Confectionary and sugar have limited PEEs available. As above, Muhammad (2011) estimated a general price elasticity for “drinks and sweets” of 0.96 in low-income countries and 0.61 in middle income countries. Snowdon estimated a PEE of 0.8—2 for confectionary, as a luxury good, and 0.5—0.7 for sugar, as a staple, in Tonga. Green and colleagues estimate a price elasticity for sweets, confectionary and sweetened beverages of 0.74 in low-income countries and 0.68 in middle income countries. Based on these, for Fiji we propose a PEE for sugar of 0.7 and a PE of 1 for confectionary and edible ices

Savoury snacks and sweet baked goods include a varied group of food products, and price elasticity is likely to depend on the degree to which substitution is possible. Estimates for price elasticity of high energy density foods are limited for LMIC. Snowdon estimated a PEE of 1—2 for fried packet snacks and cheese in Tonga. Estimates from high income countries range from 0.5 to 2.11. If less healthy snack foods are narrowly defined (such that substitution to healthier products is possible), we propose a PEE for savoury snacks in Fiji of 1.5.

While savoury and sweet breakfast biscuits and packet noodles are generally viewed nutritionally as snacks or convenience foods, in Fiji they are frequently eaten for breakfast and lunch. Cereals and staples are generally fairly price inelastic, and our review found PEEs for cereals between 0.01 to 0.61, or 0.18 to 0.61 in LMIC. While the price elasticity for cereals is generally low, in Fiji there is substantial opportunity for substitution (i.e. bread, baked goods and rice) thus a PE of 0.8, 1 and 1.5 was estimated for packaged noodles, sweet biscuits and staple biscuits, respectively..

Logic models and assumptions

Logic models for the impact of the tax

We developed logic models to explain the basis of the models estimating impact of the tax on consumption and health, revenue and household expenditure.

Consumption and health: Tax on production and import -> increased cost to consumers -> decreased purchase -> some level of substitution -> decreased consumption -> decreased intake of fat/salt/sugar - > change to NCD risk/prevalence

Based on current average prices of goods (at a category level) we also calculate the potential impact on revenue and household income.

32 Part 4: Modeling of fiscal policy impact

Tax on production and import -> increased government revenue -> increased cost of goods to consumers -> consumer response to price increase (change to consumption using price elasticity estimate) -> change in household food expenditure

Assumptions for modelling We based our assumptions for the modelling of the impact of the taxes on consumption on the recent systematic review by Thow and colleagues (Thow et al, 2014) and also on the approach to modelling the effects of interventions, including taxation, by Snowdon and colleagues (Snowden, 2010).

For all models, we assume the following 1. Tax/tax removal is fully passed to consumers 2. Consumers substitute with untaxed goods 3. Nothing else changes 4. Price control changes to allow price increases or decreases 5. Tourist consumption is negligible 6. Policy change is sustained

When modelling the impact on revenue, we assumed: 1. Consumers substituted taxed products with untaxed products (ie. substitution would not contribute to revenue) 2. Consumers aged 0—14 years and >65 years consumed 60 percent of a full 'average adult' equivalent (informed by consultation with the National Food and Nutrition Centre) When modelling the impact on household expenditure, we assumed: 1. Individual and household expenditure would be decreased by the cost of the taxed good (proportional to the estimated decrease in consumption) 2. We estimated that the average household comprised 4 adult-equivalents, based on the HIES finding that the average household was equal to 5 persons (we have dietary data available for the average “adult-equivalent). Our estimate of 4 adult- equivalents per household was determined in consultation with a nutritional statistician in Fiji, based on the likely combinations of adults and children in Fijian households (eg. 3 adults, 2 children or 2 adults, 3 children) with an average intake of discretionary food. As for Assumptions on the impact on revenue, children 0-14 years and adults >65 years consumed 60 percent of the full average adult-equivalent.

Ranges for estimates of impact

For the modelling of impact on consumption and household expenditure, we provide a range of estimates of changes in nutrient consumption and expenditure effects based on different possible consumer responses regarding substitution: 1. High estimate: consumers make no substitution for their decreased consumption 2. Mid-estimate 1: consumers compensate for decreased purchase/consumption of less healthy foods by increasing their consumption of comparable healthier foods by the

33 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

same volume (so, for example, replacing 50g of savoury snacks with high salt and fat, with 50g of bread, or replacing sweet snacks with fruit) 3. Mid-estimate 2: consumers substitute their decreased purchase/consumption of less healthy foods by increasing their consumption of other foods by 50 percent of this volume, such that they are replacing foods with higher levels of nutrients of concern with foods containing the average nutrient density of population dietary intake for these nutrients (so, for example, replacing 50g of savoury snacks with high salt and fat, with 25g of foods containing the average salt/fat density of population dietary intake) 4. Low estimate: consumers compensate for decreased purchase/consumption of less healthy foods by increasing their consumption of other foods by the same volume, such that they are replacing foods with higher levels of nutrients of concern with foods containing the average nutrient density of population dietary intake for these nutrients (so, for example, replacing 50g of savoury snacks with high salt and fat, with 50g of foods containing the average salt/fat density of population dietary intake)

Estimated impact on consumption

We modelled the expected impact of a specific tax equal to an increase in the average price of 20 percent, 50 percent and 100 percent (20 percent being the minimum recommended in the literature for a meaningful effect on consumption) for each discretionary food category identified in Phase 4: confectionary (including sugar), cakes and sweet biscuits, savoury snacks, juices and sugar-sweetened beverages (excluding koko), edible ices, and instant noodles.

The estimated decrease in calorie consumption from the minimum recommended 20 percent tax ranged from 82 (4 percent) kcal/person/day with no substitution, to 39 (2 percent) kcal/person/day with 100 percent of the decreased volume substituted with foods at the average energy density of the diet (Figure 1). The estimated decreased fat consumption ranged from 4 (7 percent) to 3 (5 percent) grams/person/day and salt consumption decreased ranged from 122 (6 percent) to 78 (4 percent) mg/person/day (See Summary Table).

Figure 1: Percent (%) change in total dietary consumption of nutrients of concern, across tax scenarios (20%, 50% and 100% taxes) and substitutions

34 Part 4: Modeling of fiscal policy impact

Notes: (a) No substitution (high estimate) (b) Healthy substitution (mid-estimate 1) (c) 50 percent replacement with average diet (mid-estimate 2) and (d) 100 percent replacement with average diet (low estimate). Source: Authors’ calculations based on consumption data from 2014 National Nutrition Survey. See Detailed Summary Table on p.45 for data by food category.

The estimated decreased calorie consumption from the 50 percent tax ranged from 205 (10 percent) kcal/person/day with no substitution, to 98 (5 percent) kcal/person/day with 100 percent of the decreased volume substituted with foods at the average energy density of the diet (Figure 1). The estimated decreased fat consumption ranged from 11 (17 percent) to 7 (12 percent) grams/person/day and salt consumption decreases ranged from 305 (14 percent) to 194 (9 percent) mg/person/day.

The highest tax of 100 percent gave estimated decreases in calorie consumption ranging from 331 (16 percent) kcal/person/day with no substitution, to 142 (7 percent) kcal/person/day with 100 percent of the decreased volume substituted with foods at the average energy density of the diet. The estimated decreased fat consumption ranged from 15 (34 percent) to 10 (24 percent) grams/person/day and salt consumption decreases ranged from 437 (20 percent) to 239 (11 percent) mg/person/day.

Confectionary The reduction in consumption of confectionary and sugars predicted by the model for the 20 percent, 50 percent and 100 percent tax was 20 percent and 14 percent, 50 percent and 35 percent, and 100 and 75 percent, respectively. Depending on the type of substitution,

35 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji this resulted in a total decrease of between 8.5 to 5.5 calories/person/day (-2.1 to -0.2 percent), a negligible change to fat and sodium (-0.4 to 0.3 percent and 0.4 to 0.1 percent respectively). Increased salt intake is likely due to possible replacement with higher density foods in line with the average diet.

Cakes, sweet biscuits and baked goods The reduction in consumption of cakes and sweet biscuits for the 20 percent, 50 percent and 100 percent tax and a PEE of 1 (all products) was 20 percent, 50 percent and 100 percent, respectively. For cakes and biscuits, this equated to a total reduction in consumption of approximately 2.9, 7.2, and 14.5 grams/person/day, respectively. Depending on the type of substitution, this resulted in an approximate total decrease of between 45.0 and 5.3 calories/person/day (-2.2 to -0.1 percent), 1.3 to 0.2 grams of fat/person/day (-2.2 to -0.3 percent) and 34.5 to 3.0 milligrams salt/person/day (-1.6 to -0.1 percent).

Savoury snacks The reduction in consumption of savoury snacks for the 20 percent, 50 percent and 100 percent tax equated to a total reduction in consumption of approximately 11, 27 and 36 grams/person/day, respectively. Depending on the type of substitution, this resulted in an approximate total decrease of between 155 to 32.8 calories/person/day (-7 to -0.3 percent), 11 to 3 grams fat/person/day (-19 to -5 percent) and 345 to 89 milligrams salt/person/day (- 16 to -4 percent).

Beverages The estimated reduction in consumption of beverages with the 20 percent, 50 percent and 100 percent tax and a PEE of 1 (all products) equated to a total reduction in consumption of approximately 16.0, 40.0 and 80.0 millilitres/person/day, respectively. Predicted decreases in caloric intake varied from 73.3 (100 percent tax, healthy alternative) to 4.3 (20 percent tax, 50 percent replacement with average dietary density) calories/person/day. In contrast, predicted increases resulted from scenarios with 100 percent average dietary density substitution (30.3 to 6.1 calories/person/day with 100 percent tax, 20 percent tax, respectively). However, global evidence suggests that people are unlikely to substitute food for decreased beverage intake, so these are unlikely scenarios.

For our healthier substitution, we assumed the more likely scenario that consumers compensated for decreased sweetened beverage consumption by increased consumption of water and other non- caloric beverages. This scenario generates an estimated reduction in total calorie consumption of -0.70 percent for the 20 percent tax, -1.75 for the 50 percent tax and -3.50 for the 100 percent tax.

Edible ices The reduction in consumption of edible ices (ice blocks) for the 20 percent, 50 percent and 100 percent tax equated to a total reduction in consumption of approximately 0.7, 1.8 and 3.7 grams/person/day, respectively. Depending on the type of substitution, this resulted in a decrease of 5.7 — 0 calories/person/day (-0.3 — 0 percent) and a decreased fat consumption of 0.3 — 0 grams/person/day (-0.5 — 0 percent). Salt, as for beverages, was

36 Part 4: Modeling of fiscal policy impact predicted to increase with substitution at 50 percent and 100 percent of average diet density (-0.1 to 0.1 percent).

37 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

38 Part 4: Modeling of fiscal policy impact

Instant noodles The reduction in consumption of instant noodles (uncooked) for the 20 percent, 50 percent and 100 percent tax with a PEE of 0.8 was 16 percent, 40 percent and 80 percent, respectively. This equated to a total reduction in consumption of approximately 0.2, 0.4 and 0.8 grams/person/day, respectively. Depending on the type of substitution, this resulted in a decrease of between 8.8 and 1.6 calories/person/day (-0.4 — 0 percent), a decreased fat consumption of between 0.3 to negligible grams/person/day (-0.6 — 0 percent) and a decreased salt intake of between 11.4 and 2.1 milligrams/person/day (-0.5 to — 0.1 percent).

Estimated impact on health We were able to estimate the implications of these consumption changes for reductions in hypertension resulting from reduced salt intake. Our model estimates that with a 100 percent tax and no substitution, or at least substitution to a no-salt alternative that on average there will be a 0.609 gram daily reduction in salt reduction. Based on existing evidence of a linear relationship between consumption and blood pressure, such a reduction in consumption translates to around 1 mm Hg reduction in Systolic Blood Pressure (Mozaffarian et al, 2014).

Even seemingly small reductions in SBP may have significant population level effects given the recognised dose-response relationship between blood pressure reduction observed at levels above the relatively low threshold of SBP of 115 mmHg. For instance in adults aged 60-69 years a SBP reduction of 1 mm Hg has been estimated to reduce risk of stroke mortality by 4 percent and 3 percent mortality from ischaemic heart disease or other vascular causes (Lewington et al, 2002).

Evidence suggests that there are likely to be additional health benefits from reduced intakes of fat, calories, and (specifically) sugar sweetened beverages.

A meta-analysis of eight randomised controlled trials on the effect of replacing saturated fat with polyunsaturated fat on coronary heart disease, published in 2010 (Mozaffarian et al, 2010), found an overall pooled risk reduction of 19 percent (RR = 0.81, 95 percent confidence interval [CI] 0.70–0.95, p = 0.008), corresponding to 10 percent reduced CHD risk (RR = 0.90, 95 percent CI = 0.83–0.97), for each 5 percent energy of increased PUFA, without evidence for statistical heterogeneity (Q-statistic p = 0.13; I2 = 37 percent). Meta-regression identified study duration as an independent determinant of risk reduction (p = 0.017), with studies of longer duration showing greater benefits.

A meta-analysis of data nine cohorts on soft drink consumption and diabetes (Greenword et al, 2014) found that the summary relative risks for sugar-sweetened and artificially sweetened soft drinks were 1·20/330 ml per d (95 percent CI 1·12, 1·29, P< 0·001) and 1·13/330 ml per d (95 percent CI 1·02, 1·25, P= 0·02), respectively. The association with sugar-sweetened soft drinks was slightly lower in studies adjusting for BMI, consistent with BMI being involved in the causal pathway. Consistent with this finding, another meta- analysis of eight studies found that individuals in the highest category of SSB intake (around

39 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji one to two servings per day) had a 26 percent (risk ratio [RR] 1.26, 95 percent CI 1.12–1.41) greater risk of developing diabetes compared with those in the lowest category (none or less than one per month) (Hu, 2013). This review noted that experimental evidence from RCTs evaluating the effects of reducing SSBs on clinical diabetes is lacking, but that there is a strong and consistent association and a demonstrated dose–response that is likely mediated through increases in insulin resistance and chronic inflammation.

Hall and colleagues calculated a simple approximation for the effect of a reduction in caloric intake on body weight (Hall et al, 2011). Every permanent change of energy intake of 100 kJ per day (23.9 kCal) will lead to an eventual weight change of about 1 kg; it will take about 1 year to achieve half of the total weight change and 95 percent of the total weight change will result in about 3 years.

Estimated impact on revenue

We identified a specific (weight/volume based) tax rate for each category of foods equivalent to a 20 percent, 50 percent and 100 percent increase in price, based on the mean retail price per 100g of a range of foods in each category.3 For each tax scenario we calculated the estimated average annual revenue, based on the quantity consumed (new estimated consumption after tax is implemented), per person and for the total Fijian population as at 2016. As described in the Assumptions, we assumed that the non-working age population consumed proportionally less than the ‘adult equivalent’.

The annual revenue generated from the proposed 20 percent, 50 percent and 100 percent tax per person would be FJD67, FJD96, and FJD4, respectively. For the total Fijian population (898,760 as at 2016 (World Bank, 2016), the proposed 20 percent, 50 percent and 100 percent tax would equal FJD51.4 million, FJD74.3 million and FJD3.5 million, respectively (Table 9).

3 See Appendix 3 for pricing data.

40 Part 4: Modeling of fiscal policy impact

Table 9: Estimated annual revenue generated (FJD) Estimated annual revenue (FJD)

Tax applied Total

20% tax 51,441,639

50% tax 74,296,012

100% tax 3,471,354 a 60% consumption relative to average adult assumed Source: Author’s calculations, based on World Bank statistics(World Development Indicators, population data)(World DevelopmentIndicators) and price data from the National Food and Nutrition Centre

As the tax increases, it is predicted that consumption decreases, showing a trade-off between reductions in consumption and revenue generation (particularly at the 100 percent tax rate).

Estimated impact on household expenditure

We modelled the expected impact of a specific tax (equal to an increase in the average price of 20 percent, 50 percent and 100 percent) on individual and household expenditure, based on data from the Fiji Household Income and Economic Survey (HIES) (2008-09) (Fiji Bureau of Statistics (2011). We estimated that for taxes of 20 percent, 50 percent and 100 percent on discretionary foods in Fiji, an individual’s annual expenditure would be reduced by between FJD1 and FJD27, FJD73 and FJD138, and between FJD312 and 419, respectively, depending on substitution. Similarly, we estimated that for taxes of 20 percent, 50 percent and 100 percent on discretionary foods, a household’s annual expenditure would be reduced by between FJD5 and FJD109, FJD293 and FJD552, and between FJD1248 and FJD1676 respectively, depending on substitution. Depending on substitution, taxing unhealthy discretionary foods at 20 percent, 50 percent and 100 percent reduced total food expenditure by between -0.1 percent and -2.3 percent, - 6.2 and -11.7 percent and -24.0 and -35.6 percent, respectively (see Table 11). Similarly, taxing unhealthy discretionary foods at 20 percent, 50 percent and 100 percent reduced total household expenditure by between 0 percent and -0.7 percent, -2.0 and -3.7, and –8.5 percent and –11.4 percent, respectively (see Tables 10 and 11).

41 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

Table 10: Estimated impact on annual individual and total household expenditure (FJD)

Estimated change in annual total Estimated change in annual individual expenditure (FJD) total household expenditure (FJD)

Dietary Substitution 20% tax 50% tax 100% tax 20% tax 50% tax 100% tax

None -27 -138 -419 -109 -552 -1676

50%* -15 -108 -365 -61 -432 -1462

100%* -3 -78 -312 -13 -313 -1248

Healthy -1 -73 -314 -5 -293 -126

* at average diet cost Source: Author’s calculations

42 Part 4: Modeling of fiscal policy impact

Table 11 Estimated change in weekly expenditure as a percentage of average income across taxes and substitutions

20 percent tax 50 percent tax 100% percent tax

% of Dietary Weekly % of total Weekly % of % of total Weekly % of % of total total Substit- spend househol spend total food household spend total food household food tution (FJD) d spendb (FJD) spenda spendb (FJD) spenda spendb spenda

None -2.1 -2.3 -0.7 -10.6 -11.7 -3.7 -32.2 -35.6 -11.4

50%* -1.2 -1.3 -0.4 -8.3 -9.2 -2.9 -28.1 -31.0 -9.9

100%* -0.2 -0.3 -0.1 -6.0 -6.6 -2.1 -24.0 -26.5 -8.5

Healthy -0.1 -0.1 -0.0 -5.6 -6.2 -2.0 -24.2 -26.7 -8.5 a Total food expenditure = 31.9% of total household expenditure (HIES data, 2008/09). b Total household expenditure = 84.9% of average household income (HIES data, 2008/09). * % substitution, at average diet cost Source: Author’s calculations from HIES data (2008/09)

This analysis suggests that fiscal policy interventions on discretionary foods are unlikely to represent a significant financial burden to households, in the case that households reduce consumption of the taxed goods – even when households substitute with healthier foods.

43 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

Conclusion and recommendations

Taking action on nutrition and NCDs is a priority for the Government of Fiji, as articulated in its 2017 National Development Plan (Republic of Fiji, 2017. 5-Year & 20-Year National Development Plan: Transforming Fiji.). The 2017 Fiji Food Security and Nutrition Policy and Plan of Action identify the need to establish the nutrition impact, and impact on revenue, of food and beverage taxes. The population of Fiji is experiencing a triple burden of malnutrition, and high rates of obesity and NCDs. These high rates of NCDs are associated with significant social and economic costs, including lost productivity and major demands on health care services.

A significant contributor to high salt, sugar and fat intakes are discretionary foods, such as confectionary, snacks and sweet beverages. These foods contribute an average of only 9 percent of daily food intake, but a much higher proportion of nutrients of concern: 16 percent of calories, 25 percent of fat and 20 percent of salt. As a result, reductions in consumption of these products can have a substantial effect on consumptions of nutrients associated with NCD risk. Many of these foods are relatively inexpensive for the significant contribution they make to intakes of energy and nutrients of concern, and thus appropriate targets of taxation. The analysis presented here indicates the application of significant excise taxes (20-50 percent) on discretionary foods not meeting nutrient profiling criteria.

We base our recommendations for taxation on a Fiji-specific nutrient profiling model, which sets thresholds for energy, sugar, sodium, fat, and saturated fat across food categories, in line with the Fiji Dietary Guidelines (see p.35 of this report).

Globally, there is growing interest in the application of fiscal policies to create incentives for the consumption of healthier (rather than unhealthy) foods, in order to accelerate action on malnutrition in all its forms, and NCDs This is reflected in the Global Action Plan for Small Island Developing States, and the Fiji National Food Security and Nutrition Policy and Plan of Action. Internationally, the strongest evidence is for the effectiveness of targeted excise taxes on sugar sweetened beverages. There is also opportunity to increase the impact of taxation through investing revenue strategically (e.g. in Berkeley, health promotion programs designed to improve nutrition and decrease consumption of sugary drinks). The literature indicates that significant taxes, for example, 50 percent, would have a meaningful impact on diets and health.

Based on the analysis in this report, he specific discretionary food categories recommended to be taxed are:  Confectionary and sugar  Beverages (sweetened drinks, juice, milks)  Edible ices  Cakes, sweet bakery and biscuits  Savoury snacks, including instant noodles

44 Conclusion and recommendations

There is opportunity for the revenue generated by taxation to support implementation of complementary measures recommended in the Food Security and Nutrition Policy and Action Plan:  Additional incentives for reformulation, such as nutrient targets for salt, fat and sugar in processed foods, based on the nutrient profiling.  Complementary social marketing, such as a social marketing and school-based campaign that de-normalises consumption of discretionary foods would enhance the effect of the tax, promote public awareness and support for the tax, and further promote reductions in discretionary salt and sugar consumption.  Restrictions on marketing of discretionary foods, to enhance the existing the draft regulation on marketing of foods and beverages to children  Financial support for measures that target healthy food affordability and availability, such as healthy food subsidies.

The estimates for decreased consumption due to taxation of discretionary foods (with ranges due to different assumptions of substitution) were:  20 percent tax: reduction of 2 — 4 percent for calories, 5 — 7 percent for fat and 4- 6 percent for salt.  50 percent tax: reduction of 5 — 10 percent for calories, 7 — 11 percent for fat and 9 — 14 percent for salt

The decreases in intakes of nutrients of concern that are estimated here are associated with reduced risk of diabetes, cardiovascular disease and overweight. In particular, for those aged 60—69 years a SBP reduction of 1 mm Hg has been estimated to reduce risk of stroke mortality by 4 percent and 3 percent mortality from ischaemic heart disease or other vascular causes (Lewington et al, 2004).

The total annual revenue generated from the proposed 20 percent or 50 percent tax would equal $FJD51.4 and $FJD74.3 million, respectively. However, this analysis suggests that fiscal policy interventions on discretionary foods are unlikely to represent a significant financial burden to households, due to decreases in consumption of discretionary foods. Our modelling for all scenarios show either neutral effects of reductions in average household spending.

45 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

Bibliography/References

Alberti KGM, Zimmet P, Shaw J. The metabolic syndrome -- a new worldwide definition. Lancet. 2005;366 North American Edition(9491):1059-62 4p. An R, Patel D, Segal D, Sturm R. Eating better for less: a national discount program for healthy food purchases in South Africa. American Journal Health Behaviour. 2013;37(1):56–61. Anderson, I. The Economic costs of noncommunicable diseases in the Pacific Islands: A rapid stocktake of the situation in Samoa, Tonga and Vanuatu. A Health, Nutrition and Population (HNP) discussion paper for the World Bank (WB). 2013. Available from http://documents.worldbank.org/curated/en/291471468063255184/pdf/865220WP0 Econo0Box385176B000PUBLIC0.pdf [Accessed: 22 July 2017]. Andreyeva T, Chaloupka F, Brownell K 2011. Estimating the potential of taxes on sugar-sweetened beverages to reduce consumption and generate revenue. Preventive Medicine, 52(6):413-416 Andreyeva T, Long MW, Brownell KD. The impact of food prices on consumption: a systematic review of research on the price elasticity of demand for food. Am J Public Health. 2010;100:216–222. Andreyeva T, Long MW, Brownell KD. The impact of food prices on consumption: A systematic review of research on the price elasticity of demand for food. American Journal of Public Health 2010; 100(2): 216-22. Archer E, Hand GA, Blair SN. Validity of U.S. nutritional surveillance:National Health and Nutrition Examination Survey caloric energy intake data, 1971-2010. PloS one. 2013;8(10):e76632. As required by the food labelling legislation in the US and labelling legislation in Australia and New Zealand. Backholer K, Sarink D, Beauchamp A, Keating C, Loh V, Ball K, Martin J and Peeters A. The impact of a tax on sugar-sweetened beverages according to socio-economic position: a systematic review of the evidence. Public Health Nutrition, 2016. 19 (17): 3070-3084. Available from doi:10.1017/S136898001600104X [Accessed: 22 July 2017]. Basu S, Vellakkal S, Agrawal S, Stuckler D, Popkin B, Ebrahim S: Averting Obesity and Type 2 Diabetes in India through Sugar-Sweetened Beverage Taxation: An Economic-Epidemiologic Modeling Study. PLoS Med 2014, 11(1):e1001582. Basu S, Vellakkal S, Agrawal S, Stuckler D, Popkin B, Ebrahim S: Averting Obesity and Type 2 Diabetes in India through Sugar-Sweetened Beverage Taxation: An Economic-Epidemiologic Modeling Study. PLoS Med 2014, 11(1):e1001582. Baylin A, Deka R, Tuitele J, Viali S, Weeks DE, McGarvey ST. INSIG2 variants, dietary patterns and metabolic risk in Samoa. Eur J Clin Nutr. 2013;67(1):101-7. Bindon JR, Baker PT. Modernization, migration and obesity among Samoan adults. Annals Of Human Biology. 1985;12(1):67-76. Bodker M, Pisinger C, Toft U, and Jorgensen T. The Danish fat tax-Effects on consumption patterns and risk of ischaemic heart disease. Preventive Medicine 2015 77: 200-203. Briggs ADM, Mytton OT, Kehlbacher A, Tiffin R, Elhussein A, Rayner M, Jebb SA, Blakely T and Scarborough. Health impact assessment of the UK soft drinks industry levy: a comparative risk assessment modelling study. The Lancet Public Health, 2017. 2: e15-22. Available from http://dx.doi.org/10.1016/ S2468-2667(16)30037-8 [Accessed: 24 July 2017]. Cabrera Escobar M, Veerman J, Tollman S, Bertram M, Hofman K. Evidence that a tax on sugar sweetened beverages reduces the obesity rate: a meta-analysis. BMC Public Health 2013; 13(1): 1072. Carter P, Gray LJ, Troughton J, Khunti K, Davies MJ. Fruit and vegetable intake and incidence of type 2 diabetes mellitus: systematic review and meta-analysis. BMJ. 2010;341. Claro RM, Levy RB, Popkin BM, Monteiro CA: Sugar-Sweetened Beverage Taxes in Brazil. American Journal of Public Health 2012, 102(1):178-183. Coconut: Leaflet No. 16 (Revised 2012). Secretariat of the Pacific Community (SPC)

46 Appendix 1: Nutrient profiling criteria, detailed method

Colchero MA, Guerrero-Lopez CM, Molina M, and Rivera JA. Beverages sales in Mexico before and after implementation of a sugar sweetened beverage tax. PLoS One 2016. 11(9): e0163463. Cornelsen L, Green R, Turner R, et al. What happens to patterns of food consumption when food prices change? Evidence from a systematic review and meta-analysis of food price elasticities globally. Health Economics 2014: n/a-n/a. Dauchet L, Amouyel P, Hercberg S, Dallongeville J. Fruit and vegetable consumption and risk of coronary heart disease: a meta-analysis of cohort studies. J Nutr. 2006;136(10):2588-93. de Souza RJ, Mente A, Maroleanu A, Cozma AI, Ha V, Kishibe T, Uleryk E, Budylowski P, Schunemann H, Beyene J, Anand SS. Intake of saturated and trans unsaturated fatty acids and risk of all cause mortality, cardiovascular disease and type 2 diabetes: systematic review and meta- analysis of observational studies. BMJ 2015; 351: h3978. Available from doi:10.1136/bmj.h3978 [Accessed: 7 November 2017]. DiBello JR, McGarvey ST, Kraft P, Goldberg R, Campos H, Quested C, et al. Dietary Patterns Are Associated with Metabolic Syndrome in Adult Samoans. The Journal of Nutritional Epidemiology. 2009. Dignan C, Burlingame B, Kumar S, Aalbersberg W. The Pacific Islands food composition tables. In: Food and Agriculture Organization of the United Nations, editor. 2nd Ed Rome, 2004. Drewnowski A, Fulgoni Iii V. Nutrient profiling of foods: creating a nutrient-rich food index. Nutrition reviews. 2008;66(1):23-39. Euromonitor International. Cultural Variations or Peculiar Tastes: Global Prospects for Breakfast Cereals, 10 Sep 2012 Ezzati M and Riboli E. Behavioural and dietary risk factors for noncommunicable diseases. The New England Journal of Medicine, 2013. 369: 954-964. Available from DOI: 10.1056/NEJMra1203528 [Accessed: 22 July 2017]. Fable J, Thompson HR, Becker CM, Rojas N, McCulloch CE, and Madsen KA. The impact of the Berkeley excise tax on sugar-sweetened beverage consumption. American Journal Public Health 2016, 106(10): 1865-71. Fiji Bureau of Statistics (2011). Report on the 2008-2009 Household Income and Expenditure Survey (HIES) for Fiji. Available from http://prdrse4all.spc.int/system/files/2008-09_hies_report_3_0.pdf [Accessed: 22 August 2017]. Fiji Ministry of Health and Medical Services, World Health Organization. Fiji NCD Risk Factors STEPS REPORT. 2012. Food and Agriculture Organization of the United Nations, World Health Organization. Second International Conference on Nutrition. Conference Outcome Document: Rome Declaration on Nutrition. ICN2 2014/2., 2014. Food Standards Australia New Zealand (FSANZ). Nutrient reference values. 2006. Gibson, J., and Romeo, A. (2017) Fiscal-Food Policies are Likely Misinformed by Biased Price Elasticities from Household Surveys: Evidence from Melanesia. Asia & the Pacific Policy Studies, 4: 405–416. Gortmaker SL, Long MW, Resch SC, War ZJ, Cradock AL, Barrett JL, Wright DR, Sonneville KR, Giles CM, Carter RC, Moodie ML, Sacks G, Swinburn BA, Hsiao A, Vine S, Barendregt J, Vos T and Wang YC. Cost Effectiveness of Childhood Obesity Interventions: Evidence and Methods for CHOICES. American Journal of Preventive Medicine 2015, 49(1): 102-111. Government of Fiji. Customs Tariff Schedule, 2012 with 2016 Budget Amendments (Replacing Schedule 2 of the Principal Act). Green R, Cornelsen L, Dangour AD, Turner R, Shankar B, Mazzocchi M, Smith RD: The effect of rising food prices on food consumption: systematic review with meta-regression. British Medical Journal 2013, 346(f3703). Greenwood, D. C., Threapleton, D. E., Evans, C. E. et al. 2014. Association between sugar- sweetened and artificially sweetened soft drinks and type 2 diabetes: systematic review and dose– response meta-analysis of prospective studies. British Journal of Nutrition

47 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

Guerrero-Lopez CM, Unar-Munguia M, and Colchero MA. Price elasticity of the demand for soft drinks, other sugar-sweetened beverages and energy dense food in Chile. BMC Public Health 2017. 17(1): 180. Hall, K. D., G. Sacks, D. Chandramohan, C. C. Chow, Y. C. Wang, S. L. Gortmaker and B. A. Swinburn (2011). "Quantification of the effect of energy imbalance on bodyweight." The Lancet 378(9793): 826-837. Hawkes C, Smith TG, Jewell J, Wardle J, Hammond RA, Friel S, Thow AM and Kain J. Smart food policies for obesity prevention. The Lancet, 2015. 385: 2410-2421 He FJ, Li J, MacGregor GA. Effect of longer term modest salt reduction on blood pressure: Cochrane systematic review and meta-analysis of randomised trials. BMJ : British Medical Journal. 2013;346. Heart Foundation of Australia. http://heartfoundation.org.au/healthy-eating/heart-foundation-tick Australia Australia Heart Foundation; 2016 [cited 2016 March 30th ]. Heart Foundation Tick Site]. Health Promotion Research Group, Department of Public Health, University of Oxford, 2004. Hu FB, Stampfer MJ, Manson JE, Rimm E, Colditz GA, Rosner BA, et al. Dietary fat intake and the risk of coronary heart disease in women. N Engl J Med. 1997;337(21):1491-9 Hu, F. B. (2013). "Resolved: there is sufficient scientific evidence that decreasing sugar-sweetened beverage consumption will reduce the prevalence of obesity and obesity-related diseases." Obesity Reviews 14(8): 606-619 Hughes RG. Diet, food supply and obesity in the Pacific. World Health Organization, 2003. Jansen A, Parkinson S, Robertson AFS. Food and nutrition in Fiji : a historical review. Parkinson S, Jansen AAJ, Robertson AFS, editors. Suva, Fiji: Dept. of Nutrition and Dietetics, Fiji School of Medicine : Institute of Pacific Studies of University of the South Pacific; 1990. Jensen JD, Smed S, Aarup and Nielsen E. Effects of the Danish saturated fat tax on the demand for meat and dairy products. Public Health Nutrition 2016. 19(17): 3085-3094. Joint Forum Economic and Pacific Health Ministers’ Meeting. NCD Roadmap Report. A background document on preventing and controlling NCDs in the Pacific, circulated June 2014 for consideration by the, July 2014. Katan MB. Trans fatty acids and plasma lipoproteins. Nutrition reviews. 2000(6):188. Kilpatrick K. Taxing soda, saving lives. Al Jazeera, Wednesday, August 19, 2015. Available at http://projects.aljazeera.com/2015/08/mexico-obesity/soda-tax.html Lewington S, Clarke R, Qizilbash N, Peto R, Collins R; Prospective Studies Collaboration. Age-specific relevance of usual blood pressure to vascular mortality: a meta-analysis of individual data for one million adults in 61 prospective studies. Lancet. 2002 Dec 14;360(9349):1903-13. Lim S, al e. A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet. 2012;380(9859). Manyema M, Veerman LJ, Chola L, Tugendhaft A, Sartorius B, Labadarios D, Hofman KJ: The Potential Impact of a 20 percent Tax on Sugar-Sweetened Beverages on Obesity in South African Adults: A Mathematical Model. PLoS ONE 2014, 9(8):e105287. Mavoa HM and McCabe M. Sociocultural factors relating to Tongans’ and indigenous Fijians’ patterns of eating, physical activity and body size. Asia Pacific Journal of Clinical Nutrition, 2008. 17 (3): 375-384. Available from http://apjcn.nhri.org.tw/server/APJCN/17/3/375.pdf [Accessed: 34 April 2017]. Mensink RP, Zock PL, Kester ADM, Katan MB. Effects of dietary fatty acids and carbohydrates on the ratio of serum total to HDL cholesterol and on serum lipids and apolipoproteins: a meta-analysis of 60 controlled trials. The American Journal Of Clinical Nutrition. 2003;77(5):1146-55. Micha R, Wallace SK, Mozaffarian D. Red and processed meat consumption and risk of incident coronary heart disease, stroke, and diabetes mellitus: a systematic review and meta-analysis. 2010:2271. Mozaffarian D, Fahimi S, Singh GM, Micha R, Khatibzadeh S, Engell RE, Lim S, Danaei G, Ezzati M, Powles J; Global Burden of Diseases Nutrition and Chronic Diseases Expert Group. Global sodium

48 Appendix 1: Nutrient profiling criteria, detailed method consumption and death from cardiovascular causes. N Engl J Med. 2014 Aug 14;371(7):624-34. doi: 10.1056/NEJMoa1304127. Mozaffarian D. Dietary and Policy Priorities for Cardiovascular Disease, Diabetes, and Obesity. Circulation 2016; 133(2): 187. Mozaffarian D. Effects of dietary fats versus carbohydrates on coronary heart disease: a review of the evidence. Current Atherosclerosis Reports. 2005;7(6):435-45. Mozaffarian, Dariush, Micha, Renata and Wallace, Sarah. 2010. Effects on Coronary Heart Disease of Increasing Polyunsaturated Fat in Place of Saturated Fat: A Systematic Review and Meta-Analysis of Randomized Controlled Trials. PLoS Med; 7(3) National Food and Nutrition Centre. 2004 NNS data are from the 2004 Fiji National Nutrition Survey Main Report. 2007.; Suva, Fiji. National Food and Nutrition Centre. Preliminary findings from the 2014 NNS data; Suva, Fiji. National Nutrition Centre. Fiji National Nutrition Survey 2004. Suva, Fiji: 2007. Ni Mhurchu C, Eyles H, Genc M, Scarborough P, Rayner M, Mizdrak A, Noaham K, and Blakely T. Effects of health-related food taxes and subsidies on mortality from diet-related disease in New Zealand: An econometric-epidemiologic modelling study. PLoS ONE 2015, 10(7): e0128477. Ni Mhurchu C, Eyles H, Schilling C, et al. Food Prices and Consumer Demand: Differences across Income Levels and Ethnic Groups. PLoS ONE 2013; 8(10): e75934. Ninth Meeting of Ministers of Health for the Pacific Island Countries. Honiara Outcome.. 2011. Olstad, D.L., Ball, K., Abbott, G., McNaughton, S.A., Le, H.N., Ni Mhurchu, C., Pollard, C. and Crawford, D.A. 2016. A process evaluation of the Supermarket Healthy Eating for Life (SHELf) randomized controlled trial. International Journal of Behavioural Nutrition & Physical Activity, 13: 27. Cobiac LJ, Tam K, Veerman L and Blakely, T. Taxes and subsidies for improving diet and population health in Australia: A cost-effectiveness modelling study. PLoS Medicine 2017,14(2): e1002232. Ni Mhurchu C, Blakely T, Jiang Y, et al. Effects of price discounts and tailored nutrition education on supermarket purchases: a randomized controlled trial. Am J Clin Nutr. 2010;91:736–747. Paraje G: The effect of price and socio-economic level on consumption of sugar-sweetened beverages (SSB): The case of Ecuador. PLos One 2016, 30;11(3):e0152260. doi: 10.1371/journal.pone.0152260. Powell LM, Wada R, Persky JJ, Chaloupka FJ 2014. Employment Impact of Sugar-Sweetened Beverage Taxes. American Journal of Public Health, 104(4):672-677 Ravuvu A, Friel S, Thow AM, Snowdon W, Wate J. Monitoring the impact of trade agreements on national food environments: trade imports and population nutrition risks in Fiji. Globalization and Health. 2017;13(33). Rayner M, Scarborough P, Stockley L. Nutrient profiles: Options for definitions for use in relation to food promotion and children’s diets, Final report. British Heart Foundation Health Promotion Research Group, Department of Public Health, University of Oxford, 2004. Republic of Fiji. 2017. 5-Year & 20-Year National Development Plan: Transforming Fiji. Ministry of Economy, Government of Fiji. Roache, S.A. and L.O. Gostin. The Untapped Power of Soda Taxes: Incentivizing Consumers, Generating Revenue, and Altering Corporate Behavior. International Journal of Health Policy and Management, 2017. 6(9): p. 489-493. Sacks FM, Svetkey LP, Vollmer WM, Appel LJ, Bray GA, Harsha D, et al. Effects on blood pressure of reduced dietary sodium and the Dietary Approaches to Stop Hypertension (DASH) diet. DASH- Sodium Collaborative Research Group. N Engl J Med. 2001;344(1):3-10. Scarborough P, Arambepola C, Kaur A, Bhatnagar P, Rayner M. Should nutrient profile models be 'category specific' or 'across-the-board'? A comparison of the two systems using diets of British adults. Eur J Clin Nutr. 2010;64(6):553-60. Scarborough P, Rayner M, Stockley L. Developing nutrient profile models: a systematic approach. Public health nutrition. 2007;10(4):330-6.

49 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

Schwendicke F, Thomson WM, Broadbent JM and Stolpe M. Effects of Taxing Sugar-Sweetened Beverages on Caries and Treatment Costs. Journal of Dental Research 2016, 95(12): 1327-1332. Schwendicke F, Thomson WM, Broadbent JM and Stolpe M. Effects of Taxing Sugar-Sweetened Beverages on Caries and Treatment Costs. Journal of Dental Research 2016, 95(12): 1327-1332. Briggs ADM, Mytton OT, Kehlbacher A, Tiffin R, Elhussein A, Rayner M, Jebb SA, Blakely, T and Scarborough P. Health impact assessment of the UK soft drinks industry levy: a comparative risk assessment modelling study. The Lancet Public Health 2017, 2(1): e15-e22. Seale J, Regmi A, Bernstein J. International evidence on food consumption patterns. Washington DC: United States Department of Agriculture, Economic Research Service, 2003. Smed S, Scarborough P, Rayner M, and Jensen JD. The effects of the Danish saturated fat tax on food and nutrient intake and modelled health outcomes: an econometric and comparative risk assessment evaluation. European Journal of Clinical Nutrition 2016. 70(6): 681-6. Snowdon W, Moodie M, Schultz J, Swinburn B. Modelling of potential food policy interventions in Fiji and Tonga and their impacts on noncommunicable disease mortality. Food Policy 2011; 36(5): 597-605. Snowdon W, Raj A, Reeve E, Guerrero RL, Fesaitu J, Cateine K, et al. Processed foods available in the Pacific Islands. Globalization and Health 2013;9(53). Snowdon, W. (2010). Guiding Policy Change to Reduce Diet-Related Disease in the Pacific. PhD Thesis. Melbourne, Australia, Deakin University. Snowdon, W., J. Negin and M. Moodie (2011). "Building an economic case for food interventions in the Pacific." Pacific Health Dialogue 16(2): 24- 29 Subar AF, Freedman LS, Tooze JA, Kirkpatrick SI, Boushey C, Neuhouser ML, et al. Addressing Current Criticism Regarding the Value of Self-Report Dietary Data. J Nutr. 2015;145(12):2639-45. Swinburn BA, Sacks G, Hall KD, McPherson K, Finegood DT, Moodie ML, Gortmakr, SL. The global obesity pandemic: shaped by global drivers and local environments. The Lancet. 2011, 378: 804-814. Tenth Pacific Health Ministers Meeting. Apia Communiqué on Healthy Islands, NCDs and the Post- 2015 Development Agenda. 4 July 2013 Thow AM, Downs S and Jan S. A systematic review of the effectiveness of food taxes and subsidies to improve diets: Understanding the recent evidence. Nutrition Reviews 2014. 72(9): 551-565. Alagiyawanna A, Townsend N, Mytton O, Scarborough P, Roberts N, Rayner M: Studying the consumption and health outcomes of fiscal interventions (taxes and subsidies) on food and beverages in countries of different income classifications; a systematic review. BMC Public Health 2015, 15(1):887 Thow AM, Downs S, Jan S: A systematic review of the effectiveness of food taxes and subsidies to improve diets: Understanding the recent evidence. Nutrition Reviews 2014, 72(9):551-565. Thow AM, Heywood P, Schultz J, Quested C, Jan S, Colagiuri S. Trade and the Nutrition Transition: Strengthening Policy for Health in the Pacific. Ecology of Food and Nutrition 2011; 50(1): 18-42. Thow AM, Quested C, Juventin L, Kun R, Khan AN, Swinburn B 2011. Taxing soft drinks in the Pacific: Implementation lessons for improving health. Health Promotion International, 26(1):55-64. Thow, A. M., B. Swinburn, S. Colagiuri, M. Diligolevu, C. Quested, P. Vivili and S. Leeder (2010). "Trade and food policy: case studies from three Pacific Island Countries." Food Policy 35: 556-564. Threapleton DE, Greenwood DC, Evans CEL, Cleghorn CL, Nykjaer C, Woodhead C, et al. Dietary fibre intake and risk of cardiovascular disease: systematic review and meta-analysis. BMJ : British Medical Journal. 2013;347. Tordoff MG, Alleva AM. Effect of drinking soda sweetened with aspartame or high-fructose corn syrup on food intake and body weight. Am J Clin Nutr. 1990;51(6):963-9. U.S. Department of Agriculture, U.S. Department of Health and Human Services. Dietary Guidelines for Americans. Washington DC: U.S Government,, 2010. Verhagen H, van den Berg H. A simple visual model to compare existing nutrient profiling schemes. Food & Nutrition Research. 2008;52:1-5.

50 Appendix 1: Nutrient profiling criteria, detailed method

Wang X, Ouyang Y, Liu J, Zhu M, Zhao G, Bao W, et al. Fruit and vegetable consumption and mortality from all causes, cardiovascular disease, and cancer: systematic review and dose-response meta-analysis of prospective cohort studies. BMJ. 2014;349:g4490. World Bank (2016). Total population, Fiji. Available from http://data.worldbank.org/indicator/SP.POP.TOTL?locations=FJ and http://wdi.worldbank.org/table/2.1 [Accessed: 22 August 2017] World Development Indicators, available at http://databank.worldbank.org/data/reports.aspx?source=world-development-indicators World Health Organisation (WHO) 2015. Using price polices to promote healthier diets. 2015. Available from: http://www.euro.who.int/__data/assets/pdf_file/0008/273662/Using-price-policies- to-promote-healthier-diets.pdf?ua=1 [Accessed: 3 July 2017]. World Health Organization, editor Report. WHO/IASO Technical Meeting: Nutrient Profiling 2010; London. World Health Organization. Diet, nutrition and the prevention of chronic diseases: report of a joint WHO/FAO expert consultation, Geneva, 28 January 1 February 2002. Geneva: World Health Organization, 2003. World Health Organization. Fiji Global School-based Student Health Survey (2015): Fact Sheet. Suva, Fiji: 2016. World Health Organization. Fiji STEPS Survey (2002): Fact Sheet. 2004. World Health Organization. Global status report on noncommunicable diseases 2010. 2011. World Health Organization. Noncommunicable Diseases: Country Profiles 2014. Geneva: 2014. World Health Organization. Prevention of cardiovascular disease : pocket guidelines for assessment and management of cardiovascular risk : (WHO/ISH cardiovascular risk prediction charts for the European Region). Geneva: World Health Organization,, 2007. World Health Organization. WHO Nutrient Profile Model for the Western Pacific Region: A tool to protect children from food marketing. Manila: 2016. World Health Organization. WHO Regional Office for Europe nutrient profile model. Copenhagen: 2015.

51 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

Appendix 1: Detailed method for developing nutrient profiling criteria

Method used for developing a food classification system for Fiji In this section, we apply an evidence-based methodology to develop a nutrient profiling tool to improve diets and prevent NCDs in Fiji.

The purpose of this model is to provide clear criteria to identify food and beverage products which contribute to NCD risk (those high in fat, salt and sugar), which can be used as a basis to develop more targeted policy measures to reduce unhealthy consumption. The development of a nutrient profiling system needs to be sensitive to local contexts, and to address needs of consumers, regulatory agencies and the food and agricultural industries; such a system needs to provide functional and transparent assessments of relative healthfulness of foods, and support both healthier choices and reformulation.

The methodology being applied to develop a model for Fiji is informed by a review of the literature on development of nutrient profiling systems, in particular a systematic approach recommended by Rayner and Scarborough (Scarborough et al, 2007) and then used by the World Health Organization to develop models in the European (WHO, Regional Office for Europe nutrient profile model) and Western Pacific Regions (WHO, Nutrient Profile Model for the Western Pacific Region).

This method requires evidence-informed decision-making across the following 5 steps (Rayner et al, 2004): A. Selection of appropriate index nutrients, in relation to desired health outcomes; B. Consideration of the most appropriate choice of base (e.g. per 100 g, per 100 kJ); C. Identification of most appropriate type of model, in relation to intended use of model; D. Identification of the most appropriate and relevant nutrient thresholds, in relation to desired health outcomes; E. Validation and testing of the model against an objective measure of a healthy diet (in this case, food-based dietary recommendations for Fiji and the Pacific region), and in relation to representation of ‘healthy’ and ‘less healthy’ foods.

Key data sources for analysis of nutrient consumption, availability and price

Dietary Intakes To identify which foods contribute most to the high intake levels of nutrients identified as contributing most to NCDs in Fiji – sodium, fat and sugar – it is necessary to establish a baseline of individual nutrient consumption.

Consumption was examined using the results of Fiji’s National Nutrition Survey, conducted between October 2014 and March 2015. The survey involving 6185 individuals (1 percent of Fiji’s population) aged 5—45 years, collected nutrient intake data via 24-hour recall.

52 Appendix 1: Nutrient profiling criteria, detailed method

Statisticians from Fiji’s National Nutrition Centre analysed data from Fiji’s National Nutrition Survey to determine average consumption of 542 food items by adults aged between 12 and 45 years of age (Appendix 1). In order to determine daily nutrient intakes, each food item was first converted to calorie and micro and macro nutrient values using conversion tables. The FAO’s (2004) Pacific Island Food Composition Tables express the Calorie, micro and macronutrient quantity contained within each food product available in the Pacific Islands, per 100g serve (Dignan et al, 2004). These values are based on extensive research undertaken by Pacific regional institutions, dating back to the 1940’s. The results in the Pacific Food Composition Tables were then cross-checked against the US Food Composition Tables and nutrient composition values identified through the (2011) regional survey of packaged foods (Snowdon et al, 2013. A limitation of the Pacific tables is that they do not contain values for saturated fat or sugar.

Evidence shows that self-reported dietary data most often results in significant under- reporting. A study looking at the credibility of the United States National Health and Nutrition Survey identified that underreporting was likely to be approximately 281 and 365 kcal per day for men and women. Disparities between reported and likely consumption is suggested to be much greater in obese participants, well over 700kcal per day (Archer et al, 2013). For 24 recalls this problem is primarily associated with poor recall and difficulties in estimating quantities (Subar, 2015). However, dietary surveys remain useful for identifying dietary patterns and the types of food consumed.

Composition of foods available in Fiji The Pacific Food Composition Tables (cross-checked against the US Food Composition Tables) provided nutrient composition of food and beverage items. The nutrient composition of store-bought foods was sourced from a 2011 store survey conducted as a part of a regional study of packaged foods.7

While the Pacific Food Composition Tables provide nutrient data for most local foods, they do not include values for sugar and saturated fat content.

Recommended nutrient intake values

Nutrient intake recommendations specific to the Fijian population are not available. For this project, we drew on Nutrient Reference Values (NRVs; recommended intakes or upper limits for all nutrients associated with human health) developed by Food Standards Australia New Zealand (FSANZ, Nutrient reference values 2006) and nutrient intake recommendations developed by the United States of America Department of Agriculture (USDA), appropriate to their multi-racial population (U.S. Department of Health and Human Services. Dietary Guidelines for Americans). These were used as a reference point for assessing nutrient consumption as they had undergone rigorous standards development process. Recommendations from World Health Organization papers also provided nutrient requirements and dietary guidelines for fat, sodium, saturated fat, trans fat and sugar (given as a proportion of daily energy intake).

Step A: Selection of appropriate index nutrients, in relation to desired health outcomes

53 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

To address the dual concerns of problem foods and problem diets, we identified nutrients to include in the nutrient profiling model by assessing: • Those foods with a demonstrated association with health conditions of concern among the target population, based on a review of the international peer-reviewed literature regarding correlations between diets, nutrients and NCD risk • Foods available in Fiji containing high levels of ‘nutrients of concern’ using data from the 2011 Pacific store survey (Snowden, et al 2013) and Pacific Food Composition Tables. • Nutrients being consumed in amounts that are more than those recommended for good health • Foods contributing most to high intakes of fat, salt and sugar by Fijians • Foods contributing most to intake of beneficial nutrients by Fijians (Protein, fibre, Vitamin A, Iron, Calcium)

To ascertain which nutrients were being consumed in amounts that are more than those recommended for good health, we used data from the 2014 National Nutrition Survey. Each of the 534 foods included in the survey were grouped into like food categories, with classification aligned initially those in the Nutrient Profiling Model produced by the European Office of the World Health Organization (WHO, Regional Office for Europe nutrient profile model) (Appendix 2).

Groups were divided into subgroups based on their makeup, food variety and type/degree of processing. For example, staples were grouped into root crops and starches, bread and flour, rice and grains and pasta.

This analysis also identified foods acting as a significant source of beneficial nutrients for Fijians (Protein, fibre, Vitamin A, Iron, Calcium). Nutrient intakes were compared to the Nutrient Reference Values (NRVs) developed by FSANZ (Nutrient reference values, 2006) and recommendations from the United States of America Department of Agriculture (USDA, Dietary Guidelines for Americans, 2010). We have used these as a reference point for assessing nutrient consumption due to Fiji’s ongoing trade linkages with these countries (Ravuvu, 2017).

Step B: Consideration of the most appropriate choice of base (e.g. per 100g, per 100kJ)

A base was selected based on food labelling standards in Fiji, with consideration of the dominating source of origin of food imports.

Countries have adopted different bases (e.g. amount of nutrients per serving or per 100g) for food labelling; in the United States nutritional values on food products are presented as a percentage per serve where for Australia and New Zealand values are presented per 100g .4 The use of a ‘per 100g’ base is common because it is simple to conceptualise and easy to compare foods within categories. The main disadvantage of using per 100g as a base is that some foods are eaten in very small portions (e.g. butter) while others are eaten in very large quantities (e.g. sweetened beverages). In contrast, using ‘per serve’ recognises that portions

4 As required by the food labelling legislation in the US and labelling legislation in Australia and New Zealand

54 Appendix 1: Nutrient profiling criteria, detailed method of foods vary widely across foods and that if eaten in large amounts they supply more nutrients than if eaten in smaller amounts. The challenge however in relying on ‘per serve’ thresholds is that recommended serving sizes are largely determined by the food producer and vary considerably.

Step C: Identification of most appropriate type of model, in relation to intended use of model

There are two main decisions regarding the type of nutrient profiling model that is most appropriate for a given situation. The first is whether the model is threshold-based or continuous, and the second is whether the nutrient profiling model is category-specific or across-the-board.

Continuous and Threshold-based Models Continuous models assign ‘points’ from different threshold bands across a number of different nutrients in a food. Those points are added to produce a summative score regarding relative healthfulness based on their nutrient composition, and thus foods can fall along a continuous range of scores to determine whether one food is nutritionally better than another (Rayner et al, 2004).

The advantage of continuous models is that they can differentiate between foods within categories, for instance where two cereals might be accepted as ‘low in fat’ in a threshold model, continuous models can differentiate which is healthier out of the two by factoring in levels of fat, as well as other nutrients, like protein, fibre and fruit content, as well as the proportion of those nutrients, to give the cereals a clear ranking.

Continuous models are also more accurate in that there are fewer anomalies, where a food might meet a threshold but not actually be considered a ‘healthy’ food for other reasons (for example, a highly refined sweet biscuit that contains little fat or sodium should not be labelled as ‘healthy’ because biscuits are generally not very nutritious (Rayner et al, 2004). However, continuous models are complex to administer, as a score needs to be calculated for foods and the assessment of healthfulness requires a comparison between the scores for different foods. Australia’s ‘Health Star Rating’ system is an example of a continuous model.

A Threshold-based Model uses pre-defined nutrient thresholds to classify foods as ‘healthier’ or ‘less healthy’. Threshold-based models can apply multiple thresholds to a food, which might all need to be met for a food to be classed as ‘healthier’ – for example, thresholds for salt, fat and sugar, or just a single threshold, for instance sugar thresholds for drinks. Threshold-based models are relatively simple to use, because a given food either meets the threshold or doesn’t, and there is thus no calculation or comparison involved.19 The Heart Foundation’s Tick program is an example of a threshold-based model; it provides the Tick only to foods that have a sugar, fat and sodium content under certain thresholds.5

Category-Specific and Across-the-Board models

5 Heart Foundation of Australia. http://heartfoundation.org.au/healthy-eating/heart-foundation-tick Australia Australia Heart Foundation; 2016 [cited 2016 March 30th ]. Heart Foundation Tick Site].

55 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

Category-specific models assess the healthfulness of foods within defined food groups, whereas across-the-board models assess the healthfulness of all foods using one given standard. In deciding whether to use one nutrient criteria applicable to all foods/beverages (‘across-the-board’) or nutrient criteria specific to different food categories (‘category- specific’), an important factor to consider is whether the model is seeking to support dietary change across food groups (for example, from fatty meats to vegetables) or dietary change within food groups (for example, from fatty meats to less fatty meats) (Scarborough et al, 2010).

For instance, an ‘across-the-board’ model might specify that all foods with less than 10g fat per 100g be classed as ‘low fat’, where a ‘category specific’ model might change the threshold depending on the food category, for instance by making the allowable fat content of a ‘healthier’ biscuit different to the allowable fat threshold of a ‘healthier’ starchy staple. In the first example, it would be easier to see that biscuits generally have more fat than starchy staples, whereas in the second, it would be easier to identify healthier biscuits rather than less healthy biscuits, and healthier staples rather than less healthy staples.

The main advantage of using category-specific nutrient criteria over across-the-board is that the nutrient thresholds can be tailored to the food type, allowing for better targeting of ‘problem’ nutrients in specific food categories. For instance, it is perfectly acceptable that cooking oils have a higher fat content than breakfast cereals, and for breakfast cereals it is more important to use a criterion that includes sodium, fibre and sugar. Additionally this type of model might incentivise manufacturers to reformulate food in order to meet a threshold.

A disadvantage of using a category-specific model over an ‘across-the-board’ is that category specific might oversimplify classification of what is ‘healthy’, and encourage people to substitute within categories even though a particular category may be unnecessary as part of a healthy diet (i.e. if a consumer uses it to switch between different brands of tinned meats when ideally they would switch to less processed sources of protein).

Another disadvantage is that the categorisation of foods may not be clear or reflect composition of the food, for instance composite foods (e.g. curry, pepe keke and meat pies) comprise of foods from more than one group. Additionally the assignment of foods into categories is culture dependant and can vary widely. In 2008, the Dutch used 14 food categories, the Swedish Government 26 food categories and the Eurofood System used 33 (Drewnowski and Fulgoni, 2008).

Table 12: Summary of types of models for nutrient profiling

Continuous Threshold-based

Category- Assesses foods within food groups Assesses foods within food groups (categories) specific (categories) using a calculated score based based on whether they meet prescribed on nutrient composition thresholds for specific nutrients e.g. FSANZ Nutrient Profiling Scoring e.g. Australian Heart Foundation’s ‘Pick the Criterion for the Australian Health Star Tick’b Rating systema

56 Appendix 1: Nutrient profiling criteria, detailed method

Across the Assesses all foods using a standard Assesses all foods based on whether they meet board calculated score based on nutrient a standard prescribed threshold for specific composition nutrients e.g. UK OfCom Model used to underpin the e.g. Codex Alimentarius Commission’s nutrient marketing of foods to children regulations c claims thresholds used to inform what foods are allowed to make nutrient claims

a Nutrition, health and related claims, F2016C00082 (2016). b Heart Foundation of Australia. http://heartfoundation.org.au/healthy-eating/heart-foundation-tick Australia Heart Foundation; 2016 [cited 2016 March 30th ]. [Heart Foundation Tick Site]. c Rayner M, Scarborough P, Lobstein T. The UK Ofcom Nutrient Profiling Model: Defining ‘healthy’ and ‘unhealthy’ foods and drinks for TV advertising to children. 2009.

An example of a model which is category specific and continuous is the new Nutrient Profiling Scoring Criterion (NPSC) developed by FSANZ (FSANZ, Nutrient reference values). The NPSC takes into account energy, saturated fat, sodium and sugar content of food along with components such as fruit and vegetables, dietary fibre and protein; allocating points based on the nutritional composition per 100 g or 100 mL. ‘Baseline points’ are allocated for energy, saturated fat, total sugar and sodium content of the food and then ‘Modifying points’ are obtained for the percentage of the food that is fruit, vegetables, nuts and legumes including coconut. Some foods also score further modifying points for fibre and protein content.

An example of a model that is across-the-board and continuous is the UK’s OfCom Model. The model uses a similar methodology as the NPSC to generate a final single score which determines whether the food can be advertised to children. Although it is across-the board, there is one thresholds set for what is acceptable for food and another for beverages. An example of a model which is threshold-based and across-the-board is the Codex Alimentarius Commission’s nutrient claims thresholds, where global thresholds have been set for nutrient claims on all foods and drinks (such as ‘low in salt’) and these aren’t specific to food categories.

The nutrient profiling model underlying the Australian Heart Foundation’s ‘Pick the Tick’ scheme and the WHO EURO model are both threshold-based and category specific. In both schemes, foods have been categorised and each category has its own specific nutrient criteria to meet one or more thresholds. In the WHO EURO model for example, all foods in category 1 (chocolate and confectionary) are subject to nutrient thresholds for sugar and fat, where foods in category 4 (breakfast cereals) are subject to a different set of nutrient thresholds across fat, sugar and salt.

Step D: Identification of the most appropriate and relevant nutrient thresholds, in relation to desired health outcomes

The key considerations in determining nutrient criteria for a nutrient profiling model relate to the desired outcome in terms of designating foods as ‘healthy’ or ‘less healthy (Rayner et el, 2004)’. First, whether the aim is to identify ‘healthier’ foods – for example, the heart

57 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji foundation ‘tick’ program – or ‘less healthy’ foods – for example, in order to apply restrictions on marketing as the WHO Euro model is designed to do.

Second, the level of specificity required of the model. For example, whether the aim is to reduce population intake in line with specific targets (for example, the global NCD target to reduce sodium intake by 30 percent), or more general reductions in consumption of foods that contribute substantially to intakes of unhealthy nutrients (and/or to increase consumption of healthy foods).

Once the aim and purpose of the model is established, approaches to determining the nutrient criteria for a model are to either a) calculate appropriate criteria based on detailed dietary intake for the relevant population (i.e. identifying required reductions in specific nutrient intakes and calculating the reductions in food intakes required to achieve this); or b) adapt the criteria used in a relevant model (i.e. with similar aims) to the target population.

The type of modelling required to develop nutrient thresholds requires detailed dietary intake data (from a nutrition survey) for comparison against NRVs. This is used to determine the desired percent level of decrease per nutrient per person (i.e. 30 percent reduction in Sodium Intake), and that percentage of decrease is then applied to average nutrient contents for each food type to determine a ‘threshold’ for that food type. The Tripartite Classification Model from Denmark was developed using this method, labelling foods as being either ‘preferable’, ‘middle course’ or ‘exceptionally’ (meaning unhealthy). The Scheme was for use in government regulation on food claims, for consumer education, and to encourage food industry reformulation.

The other method for developing thresholds is to select a nutrient model that has been developed, tested and validated by a respected source. For example, the Codex Alimentarius Commission have defined thresholds used for declaring foods ‘high’ or ‘low’ in particular nutrients, and the Australian Heart Foundation has developed thresholds to distinguish ‘healthier’ from ‘less healthy’ foods within categories.

However, challenges associated with selecting thresholds from previous models include the lack of disclosure relating to selection of thresholds, a misalignment in purpose, and the lack of thresholds across all food categories. For example, the Heart Foundation’s Pick the Tick Program developed nutrient thresholds across 12 categories and 72 subcategories, but only publicly disclosed thresholds for 5 categories.

In terms of purpose, the Pick the Tick thresholds are designed to identify foods which are ‘healthier’, as opposed to identifying foods which are the ‘least healthy’, as may be required for a model that underpinned policies to restrict marketing or change relative prices of healthy and less healthy foods. A model that would be more appropriate for identifying less healthy food is the WHO-EURO model. This model discloses nutrient thresholds across 12 of its 17 categories (i.e. foods that fall below the stated thresholds are allowed to be marketed), while imposing a marketing ban on the others.

58 Appendix 1: Nutrient profiling criteria, detailed method

In this study, a determination of nutrient thresholds was based on the desired outcome to designate foods as ‘less healthy’, to underpin policies to disincentivise ‘less healthy’ foods. Selection of thresholds was based on the desired level of decrease per nutrient per person necessary to reduce population intake in line with FSANZ Nutrient Reference Values (FSANZ Nutrient Reference Values). We applied that rate of decrease to the mean and median nutrient contents for each food category using the Fiji Store Food Survey and the Pacific Nutrient Composition Tables. We used other well-known nutrient profiling models including the Heart Foundation Tick6 and WHO’s nutrient profiling models to inform our thresholds (WHO, Nutrient Profile Model for the Western Pacific Region).

Step E: Validation and testing

Validation and testing of the model was undertaken with expert input from Fiji’s National Food and Nutrition Centre and nutrition experts working at WHO and the George Institute for Global Health, all 3 having previously been involved in nutrient standards development for the Region. The key criteria were contextual appropriateness, functionality and feasibility. Contextual appropriateness: Nutrition and health experts in Fiji were presented with the results of the dietary analysis and stepped through the decision-making process during the development of the model. Functionality: A document featuring all foods which do and do not meet the nutrient thresholds in the NP model was provided to experts from NNS for assessment. The list was assessed for anomalies by examining whether the model had provided an appropriate indication of the healthfulness of foods. Feasibility: The model was show to members of the Fijian Government, food industry and stakeholders to assess interpretation and ease-of-use.

6 Heart Foundation of Australia. http://heartfoundation.org.au/healthy-eating/heart-foundation-tick Australia Australia Heart Foundation; 2016 [cited 2016 March 30th ]. Heart Foundation Tick Site].

59 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

Appendix 2: Detailed tables for discretionary foods not meeting thresholds

TG 1: Chocolate and confectionary – Thresholds: <5 g sugar and <5 g fat Carbohydrates Sodium FOOD_NAME Energy (kCals) Fat (g) (g) (mg) Toffee,mixed 2111.00 91.23 321.71 1536.53 Chocolate,milk 473.95 24.88 56.29 81.72 Chocolate bar,`bounty bar' 417.83 21.47 49.93 124.58 Chocolate,dark 211.65 11.56 25.38 22.30 Seed bar,fruit and nut 224.72 10.39 26.71 48.23 Peanut butter,smooth,with sal 117.01 9.56 2.83 91.39 Chocolate bar,`mars bar' 186.12 7.57 27.54 83.60 Foods above this line exceed thresholds 5 g fat Lolly,mint,confectionery 156.86 1.04 38.56 24.91 Mango,dried 27.48 0.09 6.13 0.42 Chewing gum,regular 111.99 0.09 29.08 1.80 Sweets,boiled 258.60 0.00 65.63 53.37 Syrup,golden 1496.88 0.00 377.50 655.20 Honey 679.95 0.00 171.76 29.29 Jam,unspecified 49.85 0.00 12.44 2.45 Foods above this line exceed threshold of 5 g sugar but not fat

Source: Nutrient composition data from National Nutrition Survey (2014)

60 Appendix 2: Detailed tables for discretionary foods not meeting thresholds

TG3 Beverages – Thresholds: <5 g sugar and <5 g fat Carbohydrate FOOD_NAME energyKcals Total fat (g) (g) Sodium (mg) Tang powder 1192.51 0.00 311.42 616.15 Fruit drink,orange 169.00 0.00 42.03 30.33 Cordial,blackcurrant,prepared 160.35 0.00 40.62 32.07 Soft drink,Coca-Cola 146.91 0.00 37.63 41.02 Softdrink,lemonade 149.71 0.00 37.60 59.19 Softdrink,cola 142.33 0.00 36.08 39.72 Juice,orange and mango 142.46 0.42 32.68 12.57 Juice,orange,commercial 132.18 0.00 30.84 11.02 Milkshake,other flavours,commercial 137.97 2.21 23.22 99.53 Cordial,citrus,25 % prepared 65.77 0.00 16.83 17.41 Coffee powder,instant 145.22 0.66 11.11 20.90 Milk,condensed,skim,sweet,can 184.66 0.20 39.39 86.35 Foods above this line exceed threshold of 5 g sugar Milk,condensed,whole,sweet,ca 483.22 13.46 80.01 154.80 Ovaltine powder 696.94 5.20 143.69 304.58 Chocolate,drinking,powder 557.60 9.15 112.24 381.25 Milo powder 422.85 10.25 71.68 283.20 Cocoa powder,Australian 549.87 18.11 69.74 336.39 Ovaltine powder 696.94 5.20 143.69 304.58 Cocoa beverage, cocoa powder & milk 199.04 10.66 17.68 127.36 Milo,made with milk 170.15 6.37 16.94 145.46 Foods above this line exceed threshold of 5 g fat and sugar (WHO) Wine,white 186.88 0.00 3.16 69.00 Coffee,with milk,instant & regular,ns str & amt milk 35.06 1.63 2.99 32.61 Cocoa powder,Malaysian 26.30 1.30 2.83 5.50 Tea,white,brewed from leaf/teabags,reg,ns amt milk 26.84 1.52 1.82 27.32 Coffee,brewed 5.08 0.00 1.02 5.08 Rum 2902.74 0.00 0.00 28.74 Whisky 1421.00 0.00 0.00 0.00 Tea,Indian,infused 0.00 0.00 0.00 5.33 Softdrink,cola,diet 0.00 0.00 0.00 20.58

61 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

Softdrink,lemonade,diet 0.00 0.00 0.00 85.00 Tea,lemon grass,brewed,C. col 6.68 0.67 0.00 10.01 Tea,lemon leaf,brewed,Citrus 10.59 1.06 0.00 3.53 Water,imported,bottle 0.00 0.00 0.00 78.90

Source: Nutrient composition data from National Nutrition Survey (2014)

62 Appendix 2: Detailed tables for discretionary foods not meeting thresholds

TG 3: Ices <8 g sugar and <5 g fat Energy Carbohydrates Sodium FoodID FOOD_NAME (kCals) Fat (g) (g) (mg) M020 Ice cream,vanilla 456.32 26.21 46.80 163.81 ZZ19 Ice block 64.70 0.00 15.82 5.58 Foods above this line exceed threshold of 8g sugar

Source: Nutrient composition data from National Nutrition Survey (2014)

63 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

TG5 Cakes, sweet bakery, breakfast cereals – Thresholds: <450 mg sodium <10 g sugar and <10 g fat

Energy Carbohydrates Sodium FOOD_NAME (kCals) Fat (g) (g) (mg) Corn flakes,sugar coated,Kell 1656.91 0.88 390.87 3569.40 Rice bubbles 1047.00 1.20 244.80 3180.00 Corn flakes 940.78 2.50 211.43 2977.48 Cheesecake,commercial 3580.92 231.77 315.29 2818.80 Cake,cream cake,sponge 1701.31 58.61 264.02 1365.60 Doughnut,home made 470.10 23.45 59.86 1025.86 Pikelet,plain 4950.00 8.37 92.06 629.74 Bun,fruit,glazed 462.14 6.84 84.59 475.48 Bun,fruit,glazed 462.14 6.84 84.59 475.48 Pudding,self saucing 541.62 15.61 94.10 470.48 Foods above this line exceed threshold of 450 mg sodium Cake,iced,commercial 416.36 16.63 62.40 413.01 Cake,iced,commercial 416.36 16.63 62.40 413.01 Muesli,toasted 940.50 37.35 123.75 375.75 Pie,egg custard fill, bottom crust 638.82 36.80 69.61 98.84 Pie,apple,deep,baked 599.56 34.44 66.64 309.69 Pancake,home prepared 347.18 17.54 38.48 108.27 Biscuit,cream and jam filled 357.87 17.26 48.31 155.59 Biscuit,chocolate coated 313.74 15.79 39.27 164.43 Cake,lamington 394.32 14.76 59.52 198.40 Pie, coconut 402.94 14.60 65.15 202.18 Buns, Cook Is coconut bun 604.30 14.50 108.00 386.13 Biscuit,chocolate 364.88 14.33 54.40 258.25 Cake,chocolate,home prepared 260.82 12.48 32.64 348.69 Cake,chocolate,home prepared 260.82 12.48 32.64 348.69 Biscuit,plain sweet 346.63 12.37 54.26 222.89 Cake,fruit,dark,home prepared 348.84 11.83 54.77 316.20 Cake,fruit,dark,home prepared 348.84 11.83 54.77 316.20 Biscuit,fruit-filled 420.15 10.86 75.25 175.52 Cake,banana 258.80 10.66 37.74 219.03 Foods above this line exceed threshold of 10 g total fat (but not sodium) Custard powder 178.71 0.37 44.02 168.51 Cake,plain,commercial 169.88 7.30 22.38 231.44 Cake,plain,commercial 169.88 7.30 22.38 231.44 Biscuit,shortbread 151.10 7.63 19.03 147.12 Biscuit,cream,wafer 126.18 6.72 15.66 24.23 Cake,madeleine 141.71 8.44 14.47 29.60

64 Appendix 2: Detailed tables for discretionary foods not meeting thresholds

Coco pops 42.46 0.21 39.68 74.25 Foods above this line LIKELY exceed threshold of 10 g sugar (but not sodium or fat) Rice pudding,dessert,smooth 51.42 0.84 9.31 16.50 Oats,rolled,cooked 296.07 6.26 48.40 5.69 'Weet-bix' 260.10 1.85 47.87 211.62 Oats,rolled,raw 254.73 4.00 44.70 7.41

Source: Nutrient composition data from National Nutrition Survey (2014)

65 Evidence review for nutrition-relevant pricing policies and complementary measures in Fiji

TG6 Savoury snack foods – Thresholds: <450 mg sodium and <5 g fat

Energy FOOD_NAME (kCals) Fat (g) Carbohydrates (g) Sodium (mg) Taro chips,Samoan,Leilei bran 5533.00 323.40 567.60 4785.00 Peas,dried,fried 568.66 28.74 49.50 2913.29 Popcorn,regular,commerc ial 777.55 39.61 86.52 1590.80 Taro,chips,fried 1605.20 93.82 164.67 1388.19 Scone,plain,home prepared 657.10 20.19 103.23 1323.73 Biscuit,cabin,hard,Pacific Is 798.43 15.43 144.45 1249.72 Potato crisps,plain,salted 977.22 63.32 88.57 1087.02 Cheese snack,Twistie- type 406.00 22.30 44.44 799.22 Chips,pea-flour,fried 585.08 37.07 44.39 768.26 Cheese flavour snacks,twistie 402.41 23.68 43.95 729.83 Cracker,sao,jatz 384.72 15.20 55.27 670.32 Noodles,instant,uncooke d 372.27 14.71 51.05 484.59 Cashew,roasted,salted 898.86 72.85 37.35 411.80 Foods above this line exceed threshold of 450mg sodium Island dumpling,cooked 1087.93 98.96 32.20 3.39 Nuts,mixed,salted 945.72 81.84 19.39 347.80 Peanut,kernels,roasted,sa lted 447.49 37.15 9.92 239.23 Peanut,kernel & skin,raw 340.80 28.26 5.34 0.60 Corn chips,flavoured 318.75 18.13 32.56 318.75 Peanut,kernel and skin,roaste 209.67 17.23 4.80 126.67 Noodles,Maggi- type,boiled 287.20 11.02 39.16 5.80 Biscuit,wheatmeal 313.20 11.72 45.99 291.27 Foods above this line exceed threshold of 10g fat (but not sodium)

Source: Nutrient composition data from National Nutrition Survey (2014)

66 Appendix 3: Summary of price data used in modeling

Appendix 3: Summary of price data used in modeling

Food category Product Ave price (FJD) per 100gm

Category 1 Confectionary (lollies) 1.87

— Confectionary Confectionary (chocolate) 2.83

Confectionary (spreads) 1.37

- Sugars Confectionary (other:gum) 3.54 Sugar (raw or brown) 0.17 Sugar (white) 0.47 Honey 2.28

Sugar (other: golden syrup) 1.49

Category 2 Biscuits (Sweet, plain) 0.57

- Cakes, sweet biscuits Biscuits (Shortbread) 7.00 Biscuits (Sweet, other) 1.64 Cakes (commercial eg. chocolate) 3.45 Buns (sweet, iced, fruit & glazed) 0.67 Pikelets 2.00 Cakes (cheesecake) 4.81 Pies (sweet, egg custard) 1.45

Category 3 Savoury crackers (eg. cabin) 0.57

— Savoury snacks Chips (assorted: eg. corn) 2.64 Potato crisps, plain, salted 2.22 Flavoured snacks (eg. twisties) 0.90 Popcorn, regular, commercial 2.20

Category 4 Beverages (juice) 0.40

— Beverages Beverages (drink powders e.g. tang) 2.42 Beverages (soft drinks: cola) 0.17

Beverages (other: cordial) 0.59 Beverages (fruit drink, orange) 0.35 Beverages (milk, condensed) 1.25

Category 5 — Edible ices Edible ices (Ice block) 0.81

Category 11.c — Noodles Noodles instant, uncooked) 1.07 Source: Fiji National Food and Nutrition Centre, Targeted Store Survey

67

Prices for calculating substitution costs, :

Substitute To replace (where Cost per gram (FJD) applicable)

Healthy alternatives: • Fruit • Confectionary 0.0038 (Average price) (pineapples, bananas, • Sweet biscuits/baked pawpaws, oranges) goods • Savoury snacks • Edible ices • Brown bread • Noodles 0.0020

• Porridge • Buns/pikelets 0.0044

• Water • Beverages 0.00

Source: Fiji National Food and Nutrition Centre, Targeted Store Survey

68